Arrow Research search

Author name cluster

Sheng Zhang

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

43 papers
2 author rows

Possible papers

43

AAAI Conference 2026 Conference Paper

Renormalization Group Guided Tensor Network Structure Search

  • Maolin Wang
  • Bowen Yu
  • Sheng Zhang
  • Linjie Mi
  • Wanyu Wang
  • Yiqi Wang
  • Pengyue Jia
  • Xuetao Wei

Tensor network structure search (TN-SS) aims to automatically discover optimal network topologies and rank configurations for efficient tensor decomposition in high-dimensional data representation. Despite recent advances, existing TN-SS methods face significant limitations in computational tractability, structure adaptivity, and optimization robustness across diverse tensor characteristics. They struggle with three key challenges: single-scale optimization missing multi-scale structures, discrete search spaces hindering smooth structure evolution, and separated structure-parameter optimization causing computational inefficiency. We propose RGTN (Renormalization Group guided Tensor Network search), a physics-inspired framework transforming TN-SS via multi-scale renormalization group flows. Unlike fixed-scale discrete search methods, RGTN uses dynamic scale-transformation for continuous structure evolution across resolutions. Its core innovation includes learnable edge gates for optimization-stage topology modification and intelligent proposals based on physical quantities like node tension measuring local stress, and edge information flow quantifying connectivity importance. Starting from low-complexity coarse scales and refining to finer ones, RGTN finds compact structures while escaping local minima via scale-induced perturbations. Extensive experiments on light field data, high-order synthetic tensors, and video completion tasks show RGTN achieves state-of-the-art compression ratios and runs 4-600 times faster than existing methods, validating the effectiveness of our physics-inspired approach.

PDF Details DOI

AIIM Journal 2026 Journal Article

SpineCLUE: Automatic vertebrae identification using contrastive learning and uncertainty estimation

  • Sheng Zhang
  • Hongxuan Li
  • Minheng Chen
  • Mingying Li
  • Miao Liu
  • Junxian Wu
  • Cheng Xue
  • Youyong Kong

Vertebrae identification in arbitrary fields-of-view plays a crucial role in diagnosing spine disease. Most spine CT contain only local regions, such as the neck, chest, and abdomen. Existing spine-level methods, which rely on a priori on the specific number of target vertebrae, are less able to cope with this challenge. In this paper, we propose a three-stage vertebra-level method to address the challenges in 3D CT vertebrae identification with arbitrary fields-of-view. In order to integrate contextual prior information during identification, rather than identifying independently at the vertebrae-level, we perform the vertebrae localization, segmentation and identification tasks sequentially, thus making effective use of anatomical prior information about the vertebrae throughout the process. Specifically, to improve the stability of localization and prevent failures caused by abnormal vertebral positions in 3D space, we introduce a dual-factor density clustering algorithm to acquire localization information for individual vertebrae, thereby facilitating the subsequent segmentation and recognition processes. In addition, to tackle the issue of inter-class similarity and intra-class variability, we pretrain our identification network by using a supervised contrastive learning method. To further optimize the identification results, we estimated the uncertainty of the classification network and utilized the message fusion module to combine the uncertainty scores, while aggregating global information about the spine. Our method achieves state-of-the-art results on the VerSe20 challenge benchmark.

Details DOI

IROS Conference 2025 Conference Paper

BoRe-Depth: Self-Supervised Monocular Depth Estimation with Boundary Refinement for Embedded Systems

  • Juan Li
  • Sheng Zhang
  • Chang Liu
  • Jie Li
  • Xu Zhang

Depth estimation is one of the key technologies for realizing 3D perception in unmanned systems. Monocular depth estimation has been widely researched because of its low-cost advantage, but the existing methods face the challenges of poor depth estimation performance and blurred object boundaries on embedded systems. In this paper, we propose a novel monocular depth estimation model, BoRe-Depth, which contains only 8. 7M parameters. It can accurately estimate depth maps on embedded systems and significantly improves boundary quality. Firstly, we design an Enhanced Feature Adaptive Fusion Module (EFAF) which adaptively fuses depth features to enhance boundary detail representation. Secondly, we integrate semantic knowledge into the encoder to improve the object recognition and boundary perception capabilities. Finally, BoRe-Depth is deployed on NVIDIA Jetson Orin, and runs efficiently at 50. 7 FPS. We demonstrate that the proposed model significantly outperforms previous lightweight models on multiple challenging datasets, and we provide detailed ablation studies for the proposed methods. The code is available at https://github.com/liangxiansheng093/BoRe-Depth.

Details

YNIMG Journal 2025 Journal Article

Cingulate and frontopolar cortical projections to the cerebellar vermis support prolonged reaction time in identifying negative emotional scenes in women

  • Hak Kei Wong
  • Shefali Chaudhary
  • Yu Chen
  • Jaime S. Ide
  • Sheng Zhang
  • Chiang-shan R. Li

We previously observed sex differences in the association of individual anxiety and reaction time (RT) during identification of negative emotional scenes in a Hariri task. Prolonged RT, an attention marker, in identifying negative (vs. neutral) images correlated with anxiety level in women but not in men. However, the neural circuit that supports this behavioral observation remains unclear. Here, with a larger sample (64 men and 62 women), we employed whole-brain regression on individual differences in RT during matching negative vs. neutral images or RT (negative - neutral) and evaluated the results at a corrected threshold. Women but not men showed a significant correlation between individual anxiety and RT (negative - neutral), with a slope test confirming the sex difference. In women alone the cerebellar vermis showed activity in positive correlation with RT (negative - neutral). Further, Granger causality mapping (GCM) showed multiple brain regions, including the anterior cingulate cortex/frontopolar cortex (ACC/FPC), that provide inputs to the cerebellar vermis in women. Amongst these regions, only the ACC/FPC cluster showed activity (β) in significant correlation with both STAI State score and RT (negative - neutral) in women. GCM also identified a small cluster in the pons, suggesting that the cortical pontine cerebellar circuit may support prolonged RT during identification of negative emotions. Path analyses further characterized the inter-relationships amongst the neural markers, RT, and anxiety. These findings highlight a behavioral and circuit marker of anxiety state in neurotypical women. Studies with different behavioral paradigms are needed to characterize the behavioral and neural mechanisms of male anxiety.

Details DOI

IJCAI Conference 2025 Conference Paper

DANCE: Resource-Efficient Neural Architecture Search with Data-Aware and Continuous Adaptation

  • Maolin Wang
  • Tianshuo Wei
  • Sheng Zhang
  • Ruocheng Guo
  • Wangyu Wang
  • Shanshan Ye
  • Lixin Zou
  • Xuetao Wei

Neural Architecture Search (NAS) has emerged as a powerful approach for automating neural network design. However, existing NAS methods face critical limitations in real-world deployments: architectures lack adaptability across scenarios, each deployment context requires costly separate searches, and performance consistency across diverse platforms remains challenging. We propose DANCE (Dynamic Architectures with Neural Continuous Evolution), which reformulates architecture search as a continuous evolution problem through learning distributions over architectural components. DANCE introduces three key innovations: a continuous architecture distribution enabling smooth adaptation, a unified architecture space with learned selection gates for efficient sampling, and a multi-stage training strategy for effective deployment optimization. Extensive experiments across five datasets demonstrate DANCE's effectiveness. Our method consistently outperforms state-of-the-art NAS approaches in terms of accuracy while significantly reducing search costs. Under varying computational constraints, DANCE maintains robust performance while smoothly adapting architectures to different hardware requirements. The code and appendix can be found at https: //github. com/Applied-Machine-Learning-Lab/DANCE.

PDF Details DOI

AAAI Conference 2025 Conference Paper

FlexiTex: Enhancing Texture Generation via Visual Guidance

  • Dadong Jiang
  • Xianghui Yang
  • Zibo Zhao
  • Sheng Zhang
  • Jiaao Yu
  • Zeqiang Lai
  • Shaoxiong Yang
  • Chunchao Guo

Recent texture generation methods achieve impressive results due to the powerful generative prior they leverage from large-scale text-to-image diffusion models. However, abstract textual prompts are limited in providing global textural or shape information, which results in the texture generation methods producing blurry or inconsistent patterns. To tackle this, we present FlexiTex, embedding rich information via visual guidance to generate a high-quality texture. The core of FlexiTex is the Visual Guidance Enhancement module, which incorporates more specific information from visual guidance to reduce ambiguity in the text prompt and preserve high-frequency details. To further enhance the visual guidance, we introduce a Direction-Aware Adaptation module that automatically designs direction prompts based on different camera poses, avoiding the Janus problem and maintaining semantically global consistency. Benefiting from the visual guidance, FlexiTex produces quantitatively and qualitatively sound results, demonstrating its potential to advance texture generation for real-world applications.

PDF Details DOI

TMLR Journal 2025 Journal Article

Offset Unlearning for Large Language Models

  • James Y. Huang
  • Wenxuan Zhou
  • Fei Wang
  • Fred Morstatter
  • Sheng Zhang
  • Hoifung Poon
  • Muhao Chen

Despite the strong capabilities of Large Language Models (LLMs) to acquire knowledge from their training corpora, the memorization of sensitive information in the corpora such as copyrighted, biased, and private content has led to ethical and legal concerns. In response to these challenges, unlearning has emerged as a potential remedy for LLMs affected by problematic training data. However, previous unlearning techniques are either not applicable to black-box LLMs due to required access to model internal weights, or violate data protection principles by retaining sensitive data for inference-time correction. We propose $\delta$-unlearning, an offset unlearning framework for black-box LLMs. Instead of tuning the black-box LLM itself, $\delta$-unlearning learns the logit offset needed for unlearning by contrasting the logits from a pair of smaller models. Experiments demonstrate that $\delta$-unlearning can effectively unlearn target data while maintaining similar or even stronger performance on general out-of-forget-scope tasks. $\delta$-unlearning also effectively incorporates different unlearning algorithms, making our approach a versatile solution to adapting various existing unlearning algorithms to black-box LLMs.

PDF Details

AIJ Journal 2024 Journal Article

A crossword solving system based on Monte Carlo tree search

  • Jingping Liu
  • Lihan Chen
  • Sihang Jiang
  • Chao Wang
  • Sheng Zhang
  • Jiaqing Liang
  • Yanghua Xiao
  • Rui Song

Although the development of AI in games is remarkable, intelligent machines still lag behind humans in games that require the ability of language understanding. In this paper, we focus on the crossword puzzle resolution task. Solving crossword puzzles is a challenging task since it requires the ability to answer natural language questions with knowledge and the ability to execute a search over possible answers to find an optimal set of solutions for the grid. Previous solutions are devoted to exploiting heuristic strategies in search to find solutions while having limited ability to explore the search space. We build a comprehensive system for crossword puzzle resolution based on Monte Carlo Tree Search (MCTS). As far as we know, we are the first to model the crossword puzzle resolution problem as a Markov Decision Process and apply the MCTS to solve it. We construct a dataset for crossword puzzle resolution based on daily puzzles from The New York Times with detailed specifications of both the puzzle and clue database selection. Our method achieves state-of-the-art performance on the dataset. The code of the system and experiments in this paper is publicly available: https: //www. github. com/lhlclhl/CP.

Details DOI

IROS Conference 2024 Conference Paper

MonoPlane: Exploiting Monocular Geometric Cues for Generalizable 3D Plane Reconstruction

  • Wang Zhao 0001
  • Jiachen Liu
  • Sheng Zhang
  • Yishu Li
  • Sili Chen
  • Sharon X. Huang
  • Yong-Jin Liu 0001
  • Hengkai Guo

This paper presents a generalizable 3D plane detection and reconstruction framework named MonoPlane. Unlike previous robust estimator-based works (which require multiple images or RGB-D input) and learning-based works (which suffer from domain shift), MonoPlane combines the best of two worlds and establishes a plane reconstruction pipeline based on monocular geometric cues, resulting in accurate, robust and scalable 3D plane detection and reconstruction in the wild. Specifically, we first leverage large-scale pre-trained neural networks to obtain the depth and surface normals from a single image. These monocular geometric cues are then incorporated into a proximity-guided RANSAC framework to sequentially fit each plane instance. We exploit effective 3D point proximity and model such proximity via a graph within RANSAC to guide the plane fitting from noisy monocular depths, followed by image-level multi-plane joint optimization to improve the consistency among all plane instances. We further design a simple but effective pipeline to extend this single-view solution to sparse-view 3D plane reconstruction. Extensive experiments on a list of datasets demonstrate our superior zero-shot generalizability over baselines, achieving state-of-the-art plane reconstruction performance in a transferring setting. Our code is available at https://github.com/thuzhaowang/MonoPlane.

Details

AIIM Journal 2024 Journal Article

Probing perfection: The relentless art of meddling for pulmonary airway segmentation from HRCT via a human-AI collaboration based active learning method

  • Shiyi Wang
  • Yang Nan
  • Sheng Zhang
  • Federico Felder
  • Xiaodan Xing
  • Yingying Fang
  • Javier Del Ser
  • Simon L.F. Walsh

In the realm of pulmonary tracheal segmentation, the scarcity of annotated data stands as a prevalent pain point in most medical segmentation endeavors. Concurrently, most Deep Learning (DL) methodologies employed in this domain invariably grapple with other dual challenges: the inherent opacity of ‘black box’ models and the ongoing pursuit of performance enhancement. In response to these intertwined challenges, the core concept of our Human-Computer Interaction (HCI) based learning models (RS_UNet, LC_UNet, UUNet and WD_UNet) hinge on the versatile combination of diverse query strategies and an array of deep learning models. We train four HCI models based on the initial training dataset and sequentially repeat the following steps 1–4: (1) Query Strategy: Our proposed HCI models selects those samples which contribute the most additional representative information when labeled in each iteration of the query strategy (showing the names and sequence numbers of the samples to be annotated). Additionally, in this phase, the model selects the unlabeled samples with the greatest predictive disparity by calculating the Wasserstein Distance, Least Confidence, Entropy Sampling, and Random Sampling. (2) Central line correction: The selected samples in previous stage are then used for domain expert correction of the system-generated tracheal central lines in each training round. (3) Update training dataset: When domain experts are involved in each epoch of the DL model's training iterations, they update the training dataset with greater precision after each epoch, thereby enhancing the trustworthiness of the ‘black box’ DL model and improving the performance of models. (4) Model training: Proposed HCI model is trained using the updated training dataset and an enhanced version of existing UNet. Experimental results validate the effectiveness of this Human-Computer Interaction-based approaches, demonstrating that our proposed WD-UNet, LC-UNet, UUNet, RS-UNet achieve comparable or even superior performance than the state-of-the-art DL models, such as WD-UNet with only 15 %–35 % of the training data, leading to substantial reductions (65 %–85 % reduction of annotation effort) in physician annotation time.

Details DOI

AAAI Conference 2024 Conference Paper

S3A: Towards Realistic Zero-Shot Classification via Self Structural Semantic Alignment

  • Sheng Zhang
  • Muzammal Naseer
  • Guangyi Chen
  • Zhiqiang Shen
  • Salman Khan
  • Kun Zhang
  • Fahad Shahbaz Khan

Large-scale pre-trained Vision Language Models (VLMs) have proven effective for zero-shot classification. Despite the success, most traditional VLMs-based methods are restricted by the assumption of partial source supervision or ideal target vocabularies, which rarely satisfy the open-world scenario. In this paper, we aim at a more challenging setting, Realistic Zero-Shot Classification, which assumes no annotation but instead a broad vocabulary. To address the new problem, we propose the Self Structural Semantic Alignment (S3A) framework, which extracts the structural semantic information from unlabeled data while simultaneously self-learning. Our S3A framework adopts a unique Cluster-Vote-Prompt-Realign (CVPR) algorithm, which iteratively groups unlabeled data to derive structural semantics for pseudo-supervision. Our CVPR algorithm includes iterative clustering on images, voting within each cluster to identify initial class candidates from the vocabulary, generating discriminative prompts with large language models to discern confusing candidates, and realigning images and the vocabulary as structural semantic alignment. Finally, we propose to self-train the CLIP image encoder with both individual and structural semantic alignment through a teacher-student learning strategy. Our comprehensive experiments across various generic and fine-grained benchmarks demonstrate that the S3A method substantially improves over existing VLMs-based approaches, achieving a more than 15% accuracy improvement over CLIP on average. Our codes, models, and prompts are publicly released at https://github.com/sheng-eatamath/S3A.

PDF Details DOI

YNIMG Journal 2023 Journal Article

Age-related reduction in trait anxiety: Behavioral and neural evidence of automaticity in negative facial emotion processing

  • Shefali Chaudhary
  • Sheng Zhang
  • Simon Zhornitsky
  • Yu Chen
  • Herta H. Chao
  • Chiang-shan R. Li

Trait anxiety diminishes with age, which may result from age-related decline in registering salient emotional stimuli and/or enhancement in emotion regulation. We tested the hypotheses in 88 adults 21 to 85 years of age and studied with fMRI of the Hariri task. Age-related decline in stimulus registration would manifest in delayed reaction time (RT) and diminished saliency circuit activity in response to emotional vs. neutral stimuli. Enhanced control of negative emotions would manifest in diminished limbic/emotional circuit and higher prefrontal cortical (PFC) responses to negative emotion. The results showed that anxiety was negatively correlated with age. Age was associated with faster RT and diminished activation of the medial PFC, in the area of the dorsal and rostral anterior cingulate cortex (dACC/rACC) - a hub of the saliency circuit - during matching of negative but not positive vs. neutral emotional faces. A slope test confirmed the differences in the regressions. Further, age was not associated with activation of the PFC in whole-brain regression or in region-of-interest analysis of the dorsolateral PFC, an area identified from meta-analyses of the emotion regulation literature. Together, the findings fail to support either hypothesis; rather, the findings suggest age-related automaticity in processing negative emotions as a potential mechanism of diminished anxiety. Automaticity results in faster RT and diminished anterior cingulate activity in response to negative but not positive emotional stimuli. In support, analyses of psychophysiological interaction demonstrated higher dACC/rACC connectivity with the default mode network, which has been implicated in automaticity in information processing. As age increased, individuals demonstrated faster RT with higher connectivity during matching of negative vs. neutral images. Automaticity in negative emotion processing needs to be investigated as a mechanism of age-related reduction in anxiety.

Details DOI

JMLR Journal 2023 Journal Article

Distributed Community Detection in Large Networks

  • Sheng Zhang
  • Rui Song
  • Wenbin Lu
  • Ji Zhu

Community detection for large networks poses challenges due to the high computational cost as well as heterogeneous community structures. In this paper, we consider widely existing real-world networks with “grouped communities” (or “the group structure”), where nodes within grouped communities are densely connected and nodes across grouped communities are relatively loosely connected. We propose a two-step community detection approach for such networks. Firstly, we leverage modularity optimization methods to partition the network into groups, where between-group connectivity is low. Secondly, we employ the stochastic block model (SBM) or degree-corrected SBM (DCSBM) to further partition the groups into communities, allowing for varying levels of between-community connectivity. By incorporating this two-step structure, we introduce a novel divide-and-conquer algorithm that asymptotically recovers both the group structure and the community structure. Numerical studies confirm that our approach significantly reduces computational costs while achieving competitive performance. This framework provides a comprehensive solution for detecting community structures in networks with grouped communities, offering a valuable tool for various applications. [abs] [ pdf ][ bib ] &copy JMLR 2023. ( edit, beta )

PDF Details

NeurIPS Conference 2023 Conference Paper

LLaVA-Med: Training a Large Language-and-Vision Assistant for Biomedicine in One Day

  • Chunyuan Li
  • Cliff Wong
  • Sheng Zhang
  • Naoto Usuyama
  • Haotian Liu
  • Jianwei Yang
  • Tristan Naumann
  • Hoifung Poon

Conversational generative AI has demonstrated remarkable promise for empowering biomedical practitioners, but current investigations focus on unimodal text. Multimodal conversational AI has seen rapid progress by leveraging billions of image-text pairs from the public web, but such general-domain vision-language models still lack sophistication in understanding and conversing about biomedical images. In this paper, we propose a cost-efficient approach for training a vision-language conversational assistant that can answer open-ended research questions of biomedical images. The key idea is to leverage a large-scale, broad-coverage biomedical figure-caption dataset extracted from PubMed Central, use GPT-4 to self-instruct open-ended instruction-following data from the captions, and then fine-tune a large general-domain vision-language model using a novel curriculum learning method. Specifically, the model first learns to align biomedical vocabulary using the figure-caption pairs as is, then learns to master open-ended conversational semantics using GPT-4 generated instruction-following data, broadly mimicking how a layperson gradually acquires biomedical knowledge. This enables us to train a Large Language and Vision Assistant for BioMedicine (LLaVA-Med) in less than 15 hours (with eight A100s). LLaVA-Med exhibits excellent multimodal conversational capability and can follow open-ended instruction to assist with inquiries about a biomedical image. On three standard biomedical visual question answering datasets, LLaVA-Med outperforms previous supervised state-of-the-art on certain metrics. To facilitate biomedical multimodal research, we will release our instruction-following data and the LLaVA-Med model.

PDF Details

ICML Conference 2021 Conference Paper

DouZero: Mastering DouDizhu with Self-Play Deep Reinforcement Learning

  • Daochen Zha
  • Jingru Xie
  • Wenye Ma
  • Sheng Zhang
  • Xiangru Lian
  • Xia Hu 0001
  • Ji Liu

Games are abstractions of the real world, where artificial agents learn to compete and cooperate with other agents. While significant achievements have been made in various perfect- and imperfect-information games, DouDizhu (a. k. a. Fighting the Landlord), a three-player card game, is still unsolved. DouDizhu is a very challenging domain with competition, collaboration, imperfect information, large state space, and particularly a massive set of possible actions where the legal actions vary significantly from turn to turn. Unfortunately, modern reinforcement learning algorithms mainly focus on simple and small action spaces, and not surprisingly, are shown not to make satisfactory progress in DouDizhu. In this work, we propose a conceptually simple yet effective DouDizhu AI system, namely DouZero, which enhances traditional Monte-Carlo methods with deep neural networks, action encoding, and parallel actors. Starting from scratch in a single server with four GPUs, DouZero outperformed all the existing DouDizhu AI programs in days of training and was ranked the first in the Botzone leaderboard among 344 AI agents. Through building DouZero, we show that classic Monte-Carlo methods can be made to deliver strong results in a hard domain with a complex action space. The code and an online demo are released at https: //github. com/kwai/DouZero with the hope that this insight could motivate future work.

Details

NeurIPS Conference 2021 Conference Paper

Finite Sample Analysis of Average-Reward TD Learning and $Q$-Learning

  • Sheng Zhang
  • Zhe Zhang
  • Siva Theja Maguluri

The focus of this paper is on sample complexity guarantees of average-reward reinforcement learning algorithms, which are known to be more challenging to study than their discounted-reward counterparts. To the best of our knowledge, we provide the first known finite sample guarantees using both constant and diminishing step sizes of (i) average-reward TD($\lambda$) with linear function approximation for policy evaluation and (ii) average-reward $Q$-learning in the tabular setting to find the optimal policy. A major challenge is that since the value functions are agnostic to an additive constant, the corresponding Bellman operators are no longer contraction mappings under any norm. We obtain the results for TD($\lambda$) by working in an appropriately defined subspace that ensures uniqueness of the solution. For $Q$-learning, we exploit the span seminorm contractive property of the Bellman operator, and construct a novel Lyapunov function obtained by infimal convolution of a generalized Moreau envelope and the indicator function of a set.

PDF Details

YNICL Journal 2021 Journal Article

Perceived stress, self-efficacy, and the cerebral morphometric markers in binge-drinking young adults

  • Guangfei Li
  • Thang M. Le
  • Wuyi Wang
  • Simon Zhornitsky
  • Yu Chen
  • Shefali Chaudhary
  • Tan Zhu
  • Sheng Zhang

Studies have identified cerebral morphometric markers of binge drinking and implicated cortical regions in support of self-efficacy and stress regulation. However, it remains unclear how cortical structures of self-control play a role in ameliorating stress and alcohol consumption or how chronic alcohol exposure alters self-control and leads to emotional distress. We examined the data of 180 binge (131 men) and 256 non-binge (83 men) drinkers from the Human Connectome Project. We obtained data on regional cortical thickness from the HCP and derived gray matter volumes (GMVs) with voxel-based morphometry. At a corrected threshold, binge relative to non-binge drinking men showed diminished posterior cingulate cortex (PCC) thickness and dorsomedial prefrontal cortex (dmPFC) GMV. PCC thickness and dmPFC GMVs were positively and negatively correlated with self-efficacy and perceived stress, respectively, as assessed with the NIH Emotion Toolbox. Mediation and path analyses to query the inter-relationships between the neural markers and clinical variables showed a best fit of the model with daily drinks → lower PCC thickness and dmPFC GMV → lower self-efficacy → higher perceived stress in men. In contrast, binge and non-binge drinking women did not show significant differences in regional cortical thickness or GMVs. These findings suggest a pathway whereby chronic alcohol consumption alters cortical structures and self-efficacy mediates the effects of cortical structural deficits on perceived stress in men. The findings also suggest the need to investigate multimodal neural markers underlying the interplay between stress, self-control and alcohol use behavior in women.

Details DOI

YNIMG Journal 2020 Journal Article

Neural responses to negative facial emotions: Sex differences in the correlates of individual anger and fear traits

  • Guangfei Li
  • Sheng Zhang
  • Thang M. Le
  • Xiaoying Tang
  • Chiang-shan R. Li

Studies have examined sex differences in emotion processing in health and illness. However, it remains unclear how these neural processes may relate to individual differences in affective traits. We addressed this issue with a dataset of 970 subjects (508 women) curated from the Human Connectome Project. Participants were assessed with the NIH Toolbox Emotion Measures and fMRI while identifying negative facial emotion and neutral shape targets in alternating blocks. Imaging data were analyzed with published routines and the results were reported at a corrected threshold. Men scored similarly in Anger- but lower in Fear-Affect, as compared to women. Men as compared with women engaged the occipital-temporal visual cortex, retrosplenial cortex (RSC), and both anterior and posterior cingulate cortex to a greater extent during face versus shape identification. Women relative to men engaged higher activation of bilateral middle frontal cortex. In regional brain responses to face versus shape identification, men relative to women showed more significant modulations by both Anger- and Fear- Affect traits. The left RSC and right RSC/precuneus each demonstrated activities during face vs. shape identification in negative correlation with Anger- and Fear- Affect scores in men only. Anger affect was positively correlated with prolonged RT in identifying face vs. shape target in men but not women. In contrast, women relative to men showed higher Fear-Affect score and higher activation in the right middle frontal cortex, which was more strongly correlated with prolonged RT during face vs. shape identification. Together, men and women with higher Fear-Affect demonstrated lower accuracy in identifying negative facial emotion versus neutral shape target, a relationship mediated by activity of the RSC. These findings add to the literature of sex and trait individual differences in emotion processing and may help research of sex-shared and sex-specific behavioral and neural markers of emotional disorders.

Details DOI

EAAI Journal 2020 Journal Article

Predicting customer absence for automobile 4S shops: A lifecycle perspective

  • Jiawen Wang
  • Xinjun Lai
  • Sheng Zhang
  • W.M. Wang
  • Jianghang Chen

Repair and maintenance services are among the most lucrative aspects of the entire automobile business chain. However, in the context of fierce competition, customer churns have led to the bankruptcy of several 4S (sales, spare parts, services, and surveys) shops. In this regard, a six-year dataset is utilized to study customer behaviors to aid managers identify and retain valuable but potential customer churn through a customized retention solution. First, we define the absence and presence behaviors of customers and thereafter generate absence data according to customer habits; this makes it possible to treat the customer absence prediction problem as a classification problem. Second, the repeated absence and presence behaviors of customers are considered as a whole from a lifecycle perspective. A modified recurrent neural network (RNN-2L) is proposed; it is more efficient and reasonable in structure compared with traditional RNN. The time-invariant customer features and the sequential lifecycle features are handled separately; this provides a more sensible specification of the RNN structure from a behavioral interpretation perspective. Third, a customized retention solution is proposed. By comparing the proposed model with those that are conventional, it is found that the former outperforms the latter in terms of area under the curve (AUC), confusion matrix, and amount of time consumed. The proposed customized retention solution can achieve significant profit increase. This paper not only elucidates the customer relationship management in the automobile aftermarket (where the absence and presence behaviors are infrequently considered), but also presents an efficient solution to increase the predictive power of conventional machine learning models. The latter is achieved by considering behavioral and business perspectives.

Details DOI

YNIMG Journal 2020 Journal Article

The effects of age on reward magnitude processing in the monetary incentive delay task

  • Isha Dhingra
  • Sheng Zhang
  • Simon Zhornitsky
  • Thang M. Le
  • Wuyi Wang
  • Herta H. Chao
  • Ifat Levy
  • Chiang-shan R. Li

Previous studies have suggested age-related differences in reward-directed behavior and cerebral processes in support of the age effects. However, it remains unclear how age may influence the processing of reward magnitude. Here, with 54 volunteers (22–74 years of age) participating in the Monetary Incentive Delay Task (MIDT) with explicit cues ($1, ¢1, or nil) and timed response to win, we characterized brain activations during anticipation and feedback and the effects of age on these regional activations. Behaviorally, age was associated with less reaction time (RT) difference between dollar and cent trials, as a result of slower response to the dollar trials; i. e. , age was positively correlated with RT dollar – RT cent, with RT nil as a covariate. Both age and the RT difference ($1 - ¢1) were correlated with diminished activation of the right caudate head, right anterior insula, supplementary motor area (SMA)/pre-SMA, visual cortex, parahippocampal gyrus, right superior/middle frontal gyri, and left primary motor cortex during anticipation of $1 vs. ¢1 reward. Further, these regional activities mediated the age effects on RT differences. In responses to outcomes, age was associated with decreases in regional activations to dollar vs. cent loss but only because of higher age-related responses to cent losses. Together, these findings suggest age-related differences in sensitivity to the magnitude of reward. With lower cerebral responses during anticipation to win large rewards and higher responses to outcomes of small loss, aging incurs a constricted sensitivity to the magnitude of reward.

Details DOI

YNIMG Journal 2019 Journal Article

Dynamic functional connectivity during task performance and rest predicts individual differences in attention across studies

  • Angus Ho Ching Fong
  • Kwangsun Yoo
  • Monica D. Rosenberg
  • Sheng Zhang
  • Chiang-shan R. Li
  • Dustin Scheinost
  • R. Todd Constable
  • Marvin M. Chun

Dynamic functional connectivity (DFC) aims to maximize resolvable information from functional brain scans by considering temporal changes in network structure. Recent work has demonstrated that static, i. e. time-invariant resting-state and task-based FC predicts individual differences in behavior, including attention. Here, we show that DFC predicts attention performance across individuals. Sliding-window FC matrices were generated from fMRI data collected during rest and attention task performance by calculating Pearson's r between every pair of nodes of a whole-brain atlas within overlapping 10–60s time segments. Next, variance in r values across windows was taken to quantify temporal variability in the strength of each connection, resulting in a DFC connectome for each individual. In a leave-one-subject-out-cross-validation approach, partial-least-square-regression (PLSR) models were then trained to predict attention task performance from DFC matrices. Predicted and observed attention scores were significantly correlated, indicating successful out-of-sample predictions across rest and task conditions. Combining DFC and static FC features numerically improves predictions over either model alone, but the improvement was not statistically significant. Moreover, dynamic and combined models generalized to two independent data sets (participants performing the Attention Network Task and the stop-signal task). Edges with significant PLSR coefficients concentrated in visual, motor, and executive-control brain networks; moreover, most of these coefficients were negative. Thus, better attention may rely on more stable, i. e. less variable, information flow between brain regions.

Details DOI

ICML Conference 2019 Conference Paper

Greedy Orthogonal Pivoting Algorithm for Non-Negative Matrix Factorization

  • Kai Zhang
  • Sheng Zhang
  • Jun Liu
  • Jun Wang
  • Jie Zhang

Non-negative matrix factorization is a powerful tool for learning useful representations in the data and has been widely applied in many problems such as data mining and signal processing. Orthogonal NMF, which can improve the locality of decomposition, has drawn considerable interest in solving clustering problems in recent years. However, imposing simultaneous non-negative and orthogonal structure can be quite difficult, and so existing algorithms can only solve it approximately. To address this challenge, we propose an innovative procedure called Greedy Orthogonal Pivoting Algorithm (GOPA). The GOPA algorithm fully exploits the sparsity of non-negative orthogonal solutions to break the global problem into a series of local optimizations, in which an adaptive subset of coordinates are updated in a greedy, closed-form manner. The biggest advantage of GOPA is that it promotes exact orthogonality and provides solid empirical evidence that stronger orthogonality does contribute favorably to better clustering performance. On the other hand, we further design randomized and parallel version of GOPA, which can further reduce the computational cost and improve accuracy, making it suitable for large data.

Details

IJCAI Conference 2019 Conference Paper

Omnidirectional Scene Text Detection with Sequential-free Box Discretization

  • Yuliang Liu
  • Sheng Zhang
  • Lianwen Jin
  • Lele Xie
  • Yaqiang Wu
  • Zhepeng Wang

Scene text in the wild is commonly presented with high variant characteristics. Using quadrilateral bounding box to localize the text instance is nearly indispensable for detection methods. However, recent researches reveal that introducing quadrilateral bounding box for scene text detection will bring a label confusion issue which is easily overlooked, and this issue may significantly undermine the detection performance. To address this issue, in this paper, we propose a novel method called Sequential-free Box Discretization (SBD) by discretizing the bounding box into key edges (KE) which can further derive more effective methods to improve detection performance. Experiments showed that the proposed method can outperform state-of-the-art methods in many popular scene text benchmarks, including ICDAR 2015, MLT, and MSRA-TD500. Ablation study also showed that simply integrating the SBD into Mask R-CNN framework, the detection performance can be substantially improved. Furthermore, an experiment on the general object dataset HRSC2016 (multi-oriented ships) showed that our method can outperform recent state-of-the-art methods by a large margin, demonstrating its powerful generalization ability.

PDF Details

YNIMG Journal 2018 Journal Article

Connectome-based predictive modeling of attention: Comparing different functional connectivity features and prediction methods across datasets

  • Kwangsun Yoo
  • Monica D. Rosenberg
  • Wei-Ting Hsu
  • Sheng Zhang
  • Chiang-shan R. Li
  • Dustin Scheinost
  • R. Todd Constable
  • Marvin M. Chun

Connectome-based predictive modeling (CPM; Finn et al. , 2015; Shen et al. , 2017) was recently developed to predict individual differences in traits and behaviors, including fluid intelligence (Finn et al. , 2015) and sustained attention (Rosenberg et al. , 2016a), from functional brain connectivity (FC) measured with fMRI. Here, using the CPM framework, we compared the predictive power of three different measures of FC (Pearson's correlation, accordance, and discordance) and two different prediction algorithms (linear and partial least square [PLS] regression) for attention function. Accordance and discordance are recently proposed FC measures that respectively track in-phase synchronization and out-of-phase anti-correlation (Meskaldji et al. , 2015). We defined connectome-based models using task-based or resting-state FC data, and tested the effects of (1) functional connectivity measure and (2) feature-selection/prediction algorithm on individualized attention predictions. Models were internally validated in a training dataset using leave-one-subject-out cross-validation, and externally validated with three independent datasets. The training dataset included fMRI data collected while participants performed a sustained attention task and rested (N = 25; Rosenberg et al. , 2016a). The validation datasets included: 1) data collected during performance of a stop-signal task and at rest (N = 83, including 19 participants who were administered methylphenidate prior to scanning; Farr et al. , 2014a; Rosenberg et al. , 2016b), 2) data collected during Attention Network Task performance and rest (N = 41, Rosenberg et al. , in press), and 3) resting-state data and ADHD symptom severity from the ADHD-200 Consortium (N = 113; Rosenberg et al. , 2016a). Models defined using all combinations of functional connectivity measure (Pearson's correlation, accordance, and discordance) and prediction algorithm (linear and PLS regression) predicted attentional abilities, with correlations between predicted and observed measures of attention as high as 0. 9 for internal validation, and 0. 6 for external validation (all p's < 0. 05). Models trained on task data outperformed models trained on rest data. Pearson's correlation and accordance features generally showed a small numerical advantage over discordance features, while PLS regression models were usually better than linear regression models. Overall, in addition to correlation features combined with linear models (Rosenberg et al. , 2016a), it is useful to consider accordance features and PLS regression for CPM.

Details DOI

YNICL Journal 2018 Journal Article

Dynamic network dysfunction in cocaine dependence: Graph theoretical metrics and stop signal reaction time

  • Yihe Zhang
  • Sheng Zhang
  • Jaime S. Ide
  • Sien Hu
  • Simon Zhornitsky
  • Wuyi Wang
  • Guozhao Dong
  • Xiaoying Tang

Graphic theoretical metrics have become increasingly popular in characterizing functional connectivity of neural networks and how network connectivity is compromised in neuropsychiatric illnesses. Here, we add to this literature by describing dynamic network connectivities of 78 cocaine dependent (CD) and 85 non-drug using healthy control (HC) participants who underwent fMRI during performance of a stop signal task (SST). Compared to HC, CD showed prolonged stop signal reaction time (SSRT), consistent with deficits in response inhibition. In graph theoretical analysis of dynamic functional connectivity, we examined temporal flexibility and spatiotemporal diversity of 14 networks covering the whole brain. Temporal flexibility quantifies how frequently a brain region interacts with regions of other communities across time, with high temporal flexibility indicating that a region interacts predominantly with regions outside its own community. Spatiotemporal diversity quantifies how uniformly a brain region interacts with regions in other communities over time, with high spatiotemporal diversity indicating that the interactions are more evenly distributed across communities. Compared to HC, CD exhibited decreased temporal flexibility and increased spatiotemporal diversity in the great majority of neural networks. The graph metric measures of the default mode network negatively correlated with SSRT in CD but not HC. The findings are consistent with diminished temporal flexibility and a compensatory increase in spatiotemporal diversity, in association with impairment of a critical executive function, in cocaine addiction. More broadly, the findings suggest that graph theoretical metrics provide new insights for connectivity analyses to elucidate network dysfunction that may elude conventional measures.

Details DOI

AAAI Conference 2018 Conference Paper

Feature Enhancement Network: A Refined Scene Text Detector

  • Sheng Zhang
  • Yuliang Liu
  • Lianwen Jin
  • Canjie Luo

In this paper, we propose a refined scene text detector with a novel Feature Enhancement Network (FEN) for Region Proposal and Text Detection Refinement. Retrospectively, both region proposal with only 3 × 3 sliding-window feature and text detection refinement with single scale high level feature are insufficient, especially for smaller scene text. Therefore, we design a new FEN network with taskspecific, low and high level semantic features fusion to improve the performance of text detection. Besides, since unitary position-sensitive RoI pooling in general object detection is unreasonable for variable text regions, an adaptively weighted position-sensitive RoI pooling layer is devised for further enhancing the detecting accuracy. To tackle the sample-imbalance problem during the refinement stage, we also propose an effective positives mining strategy for ef- ficiently training our network. Experiments on ICDAR 2011 and 2013 robust text detection benchmarks demonstrate that our method can achieve state-of-the-art results, outperforming all reported methods in terms of F-measure.

PDF Details

IJCAI Conference 2017 Conference Paper

How to Keep a Knowledge Base Synchronized with Its Encyclopedia Source

  • Jiaqing Liang
  • Sheng Zhang
  • Yanghua Xiao

Knowledge bases are playing an increasingly important role in many real-world applications. However, most of these knowledge bases tend to be outdated, which limits the utility of these knowledge bases. In this paper, we investigate how to keep the freshness of the knowledge base by synchronizing it with its data source (usually encyclopedia websites). A direct solution is revisiting the whole encyclopedia periodically and rerun the entire pipeline of the construction of knowledge base like most existing methods. However, this solution is wasteful and incurs massive overload of the network, which limits the update frequency and leads to knowledge obsolescence. To overcome the weakness, we propose a set of synchronization principles upon which we build an Update System for knowledge Base (USB) with an update frequency predictor of entities as the core component. We also design a set of effective features and realize the predictor. We conduct extensive experiments to justify the effectiveness of the proposed system, model, as well as the underlying principles. Finally, we deploy USB on a Chinese knowledge base to improve its freshness.

PDF Details

YNICL Journal 2017 Journal Article

Sex differences in the interacting roles of impulsivity and positive alcohol expectancy in problem drinking: A structural brain imaging study

  • Jaime S. Ide
  • Simon Zhornitsky
  • Sien Hu
  • Sheng Zhang
  • John H. Krystal
  • Chiang-shan R. Li

Alcohol expectancy and impulsivity are implicated in alcohol misuse. However, how these two risk factors interact to determine problem drinking and whether men and women differ in these risk processes remain unclear. In 158 social drinkers (86 women) assessed for Alcohol Use Disorder Identification Test (AUDIT), positive alcohol expectancy, and Barratt impulsivity, we examined sex differences in these risk processes. Further, with structural brain imaging, we examined the neural bases underlying the relationship between these risk factors and problem drinking. The results of general linear modeling showed that alcohol expectancy best predicted problem drinking in women, whereas in men as well as in the combined group alcohol expectancy and impulsivity interacted to best predict problem drinking. Alcohol expectancy was associated with decreased gray matter volume (GMV) of the right posterior insula in women and the interaction of alcohol expectancy and impulsivity was associated with decreased GMV of the left thalamus in women and men combined and in men alone, albeit less significantly. These risk factors mediated the correlation between GMV and problem drinking. Conversely, models where GMV resulted from problem drinking were not supported. These new findings reveal distinct psychological factors that dispose men and women to problem drinking. Although mediation analyses did not determine a causal link, GMV reduction in the insula and thalamus may represent neural phenotype of these risk processes rather than the consequence of alcohol consumption in non-dependent social drinkers. The results add to the alcohol imaging literature which has largely focused on dependent individuals and help elucidate alterations in brain structures that may contribute to the transition from social to habitual drinking.

Details DOI

YNIMG Journal 2016 Journal Article

A dual but asymmetric role of the dorsal anterior cingulate cortex in response inhibition and switching from a non-salient to salient action

  • Peter Manza
  • Sien Hu
  • Herta H. Chao
  • Sheng Zhang
  • Hoi-Chung Leung
  • Chiang-shan R. Li

Response inhibition and salience detection are among the most studied psychological constructs of cognitive control. Despite a growing body of work, how inhibition and salience processing interact and engage regional brain activations remains unclear. Here, we examined this issue in a stop signal task (SST), where a prepotent response needs to be inhibited to allow an alternative, less dominant response. Sixteen adult individuals performed two versions of the SST each with 25% (SST25) and 75% (SST75) of stop trials. We posited that greater regional activations to the infrequent trial type in each condition (i. e. , to stop as compared to go trials in SST25 and to go as compared to stop trials in SST75) support salience detection. Further, successful inhibition in stop trials requires attention to the stop signal to trigger motor inhibition, and the stop signal reaction time (SSRT) has been used to index the efficiency of motor response inhibition. Therefore, greater regional activations to stop as compared to go success trials in association with the stop signal reaction time (SSRT) serve to expedite response inhibition. In support of an interactive role, the dorsal anterior cingulate cortex (dACC) increases activation to salience detection in both SST25 and SST75, but only mediates response inhibition in SST75. Thus, infrequency response in the dACC supports motor inhibition only when stopping has become a routine. In contrast, although the evidence is less robust, the pre-supplementary motor area (pre-SMA) increases activity to the infrequent stimulus and supports inhibition in both SST25 and SST75. These findings clarify a unique role of the dACC and add to the literature that distinguishes dACC and pre-SMA functions in cognitive control.

Details DOI

YNICL Journal 2016 Journal Article

Cocaine dependence and thalamic functional connectivity: a multivariate pattern analysis

  • Sheng Zhang
  • Sien Hu
  • Rajita Sinha
  • Marc N. Potenza
  • Robert T. Malison
  • Chiang-shan R. Li

Cocaine dependence is associated with deficits in cognitive control. Previous studies demonstrated that chronic cocaine use affects the activity and functional connectivity of the thalamus, a subcortical structure critical for cognitive functioning. However, the thalamus contains nuclei heterogeneous in functions, and it is not known how thalamic subregions contribute to cognitive dysfunctions in cocaine dependence. To address this issue, we used multivariate pattern analysis (MVPA) to examine how functional connectivity of the thalamus distinguishes 100 cocaine-dependent participants (CD) from 100 demographically matched healthy control individuals (HC). We characterized six task-related networks with independent component analysis of fMRI data of a stop signal task and employed MVPA to distinguish CD from HC on the basis of voxel-wise thalamic connectivity to the six independent components. In an unbiased model of distinct training and testing data, the analysis correctly classified 72% of subjects with leave-one-out cross-validation (p < 0.001), superior to comparison brain regions with similar voxel counts (p < 0.004, two-sample t test). Thalamic voxels that form the basis of classification aggregate in distinct subclusters, suggesting that connectivities of thalamic subnuclei distinguish CD from HC. Further, linear regressions provided suggestive evidence for a correlation of the thalamic connectivities with clinical variables and performance measures on the stop signal task. Together, these findings support thalamic circuit dysfunction in cognitive control as an important neural marker of cocaine dependence.

Details DOI

YNICL Journal 2016 Journal Article

Power spectrum scale invariance as a neural marker of cocaine misuse and altered cognitive control

  • Jaime S. Ide
  • Sien Hu
  • Sheng Zhang
  • Lilianne R. Mujica-Parodi
  • Chiang-shan R. Li

BACKGROUND: Magnetic resonance imaging (MRI) has highlighted the effects of chronic cocaine exposure on cerebral structures and functions, and implicated the prefrontal cortices in deficits of cognitive control. Recent investigations suggest power spectrum scale invariance (PSSI) of cerebral blood oxygenation level dependent (BOLD) signals as a neural marker of cerebral activity. We examined here how PSSI is altered in association with cocaine misuse and impaired cognitive control. METHODS: Eighty-eight healthy (HC) and seventy-five age and gender matched cocaine dependent (CD) adults participated in functional MRI of a stop signal task (SST). BOLD images were preprocessed using standard procedures in SPM, including detrending, band-pass filtering (0.01-0.25 Hz), and correction for head motions. Voxel-wise PSSI measures were estimated by a linear fit of the power spectrum with a log-log scale. In group analyses, we examined differences in PSSI between HC and CD, and its association with clinical and behavioral variables using a multiple regression. A critical component of cognitive control is post-signal behavioral adjustment, which is compromised in cocaine dependence. Therefore, we examined the PSSI changes in association with post-signal slowing (PSS) in the SST. RESULTS: Compared to HC, CD showed decreased PSS and PSSI in multiple frontoparietal regions. PSSI was positively correlated with PSS in HC in multiple regions, including the left inferior frontal gyrus (IFG) and right supramarginal gyrus (SMG), which showed reduced PSSI in CD. CONCLUSIONS: These findings suggest disrupted connectivity dynamics in the fronto-parietal areas in association with post-signal behavioral adjustment in cocaine addicts. These new findings support PSSI as a neural marker of impaired cognitive control in cocaine addiction.

Details DOI

YNIMG Journal 2015 Journal Article

Anticipating conflict: Neural correlates of a Bayesian belief and its motor consequence

  • Sien Hu
  • Jaime S. Ide
  • Sheng Zhang
  • Chiang-shan R. Li

Previous studies have examined the neural correlates of proactive control using a variety of behavioral paradigms; however, the neural network relating the control process to its behavioral consequence remains unclear. Here, we applied a dynamic Bayesian model to a large fMRI data set of the stop signal task to address this issue. By estimating the probability of the stop signal – p(Stop) – trial by trial, we showed that higher p(Stop) is associated with prolonged go trial reaction time (RT), indicating proactive control of motor response. In modeling fMRI signals at trial and target onsets, we distinguished activities of proactive control, prediction error, and RT slowing. We showed that the anterior pre-supplementary motor area (pre-SMA) responds specifically to increased stop signal likelihood, and its activity is correlated with activations of the posterior pre-SMA and bilateral anterior insula during prolonged response times. This directional link is also supported by Granger causality analysis. Furthermore, proactive control, prediction error, and time-on-task are each mapped to distinct areas in the medial prefrontal cortex. Together, these findings dissect regional functions of the medial prefrontal cortex in cognitive control and provide system level evidence associating conflict anticipation with its motor consequence.

Details DOI

YNICL Journal 2015 Journal Article

Conflict anticipation in alcohol dependence — A model-based fMRI study of stop signal task

  • Sien Hu
  • Jaime S. Ide
  • Sheng Zhang
  • Rajita Sinha
  • Chiang-shan R. Li

BACKGROUND: Our previous work characterized altered cerebral activations during cognitive control in individuals with alcohol dependence (AD). A hallmark of cognitive control is the ability to anticipate changes and adjust behavior accordingly. Here, we employed a Bayesian model to describe trial-by-trial anticipation of the stop signal and modeled fMRI signals of conflict anticipation in a stop signal task. Our goal is to characterize the neural correlates of conflict anticipation and its relationship to response inhibition and alcohol consumption in AD. METHODS: Twenty-four AD and 70 age and gender matched healthy control individuals (HC) participated in the study. fMRI data were pre-processed and modeled with SPM8. We modeled fMRI signals at trial onset with individual events parametrically modulated by estimated probability of the stop signal, p(Stop), and compared regional responses to conflict anticipation between AD and HC. To address the link to response inhibition, we regressed whole-brain responses to conflict anticipation against the stop signal reaction time (SSRT). RESULTS: Compared to HC (54/70), fewer AD (11/24) showed a significant sequential effect - a correlation between p(Stop) and RT during go trials - and the magnitude of sequential effect is diminished, suggesting a deficit in proactive control. Parametric analyses showed decreased learning rate and over-estimated prior mean of the stop signal in AD. In fMRI, both HC and AD responded to p(Stop) in bilateral inferior parietal cortex and anterior pre-supplementary motor area, although the magnitude of response increased in AD. In contrast, HC but not AD showed deactivation of the perigenual anterior cingulate cortex (pgACC). Furthermore, deactivation of the pgACC to increasing p(Stop) is positively correlated with the SSRT in HC but not AD. Recent alcohol consumption is correlated with increased activation of the thalamus and cerebellum in AD during conflict anticipation. CONCLUSIONS: The current results highlight altered proactive control that may serve as an additional behavioral and neural marker of alcohol dependence.

Details DOI

YNIMG Journal 2015 Journal Article

Randomized structural sparsity via constrained block subsampling for improved sensitivity of discriminative voxel identification

  • Yilun Wang
  • Junjie Zheng
  • Sheng Zhang
  • Xunjuan Duan
  • Huafu Chen

In this paper, we consider voxel selection for functional Magnetic Resonance Imaging (fMRI) brain data with the aim of finding a more complete set of probably correlated discriminative voxels, thus improving interpretation of the discovered potential biomarkers. The main difficulty in doing this is an extremely high dimensional voxel space and few training samples, resulting in unreliable feature selection. In order to deal with the difficulty, stability selection has received a great deal of attention lately, especially due to its finite sample control of false discoveries and transparent principle for choosing a proper amount of regularization. However, it fails to make explicit use of the correlation property or structural information of these discriminative features and leads to large false negative rates. In other words, many relevant but probably correlated discriminative voxels are missed. Thus, we propose a new variant on stability selection “randomized structural sparsity”, which incorporates the idea of structural sparsity. Numerical experiments demonstrate that our method can be superior in controlling for false negatives while also keeping the control of false positives inherited from stability selection.

Details DOI

YNIMG Journal 2015 Journal Article

The effects of age on resting state functional connectivity of the basal ganglia from young to middle adulthood

  • Peter Manza
  • Sheng Zhang
  • Sien Hu
  • Herta H. Chao
  • Hoi-Chung Leung
  • Chiang-shan R. Li

The basal ganglia nuclei are critical for a variety of cognitive and motor functions. Much work has shown age-related structural changes of the basal ganglia. Yet less is known about how the functional interactions of these regions with the cerebral cortex and the cerebellum change throughout the lifespan. Here, we took advantage of a convenient sample and examined resting state functional magnetic resonance imaging data from 250 adults 18 to 49years of age, focusing specifically on the caudate nucleus, pallidum, putamen, and ventral tegmental area/substantia nigra (VTA/SN). There are a few main findings to report. First, with age, caudate head connectivity increased with a large region of ventromedial prefrontal/medial orbitofrontal cortex. Second, across all subjects, pallidum and putamen showed negative connectivity with default mode network (DMN) regions such as the ventromedial prefrontal cortex and posterior cingulate cortex, in support of anti-correlation of the “task-positive” network (TPN) and DMN. This negative connectivity was reduced with age. Furthermore, pallidum, posterior putamen and VTA/SN connectivity to other TPN regions, such as somatomotor cortex, decreased with age. These results highlight a distinct effect of age on cerebral functional connectivity of the dorsal striatum and VTA/SN from young to middle adulthood and may help research investigating the etiologies or monitoring outcomes of neuropsychiatric conditions that implicate dopaminergic dysfunction.

Details DOI

AAAI Conference 2015 Conference Paper

What Is the Longest River in the USA? Semantic Parsing for Aggregation Questions

  • Kun Xu
  • Sheng Zhang
  • Yansong Feng
  • Songfang Huang
  • Dongyan Zhao

Answering natural language questions against structured knowledge bases (KB) has been attracting increasing attention in both IR and NLP communities. The task involves two main challenges: recognizing the questions’ meanings, which are then grounded to a given KB. Targeting simple factoid questions, many existing open domain semantic parsers jointly solve these two subtasks, but are usually expensive in complexity and resources. In this paper, we propose a simple pipeline framework to efficiently answer more complicated questions, especially those implying aggregation operations, e. g. , argmax, argmin. We first develop a transitionbased parsing model to recognize the KB-independent meaning representation of the user’s intention inherent in the question. Secondly, we apply a probabilistic model to map the meaning representation, including those aggregation functions, to a structured query. The experimental results showe that our method can better understand aggregation questions, outperforming the state-of-the-art methods on the Free917 dataset while still maintaining promising performance on a more challenging dataset, WebQuestions, without extra training.

PDF Details

YNICL Journal 2014 Journal Article

Error-related functional connectivity of the thalamus in cocaine dependence

  • Sheng Zhang
  • Sien Hu
  • Sarah R. Bednarski
  • Emily Erdman
  • Chiang-shan R. Li

Error processing is a critical component of cognitive control, an executive function that has been widely implicated in substance misuse. In previous studies we showed that error related activations of the thalamus predicted relapse to drug use in cocaine addicted individuals (Luo et al., 2013). Here, we investigated whether the error-related functional connectivity of the thalamus is altered in cocaine dependent patients (PCD, n = 54) as compared to demographically matched healthy individuals (HC, n = 54). The results of a generalized psychophysiological interaction analysis showed negative thalamic connectivity with the ventral medial prefrontal cortex (vmPFC), in the area of perigenual and subgenual anterior cingulate cortex, in HC but not PCD (p < 0.05, corrected, two-sample t test). This difference in functional connectivity was not observed for task-residual signals, suggesting that it is specific to task-related processes during cognitive control. Further, the thalamic-vmPFC connectivity is positively correlated with the amount of cocaine use in the prior month for female but not for male PCD. These findings add to recent literature and provide additional evidence for circuit-level biomarkers of cocaine dependence.

Details DOI

YNIMG Journal 2014 Journal Article

Resting state functional connectivity of the basal nucleus of Meynert in humans: In comparison to the ventral striatum and the effects of age

  • Chiang-shan R. Li
  • Jaime S. Ide
  • Sheng Zhang
  • Sien Hu
  • Herta H. Chao
  • Laszlo Zaborszky

The basal nucleus of Meynert (BNM) provides the primary cholinergic inputs to the cerebral cortex. Loss of neurons in the BNM is linked to cognitive deficits in Alzheimer's disease and other degenerative conditions. Numerous animal studies described cholinergic and non-cholinergic neuronal responses in the BNM; however, work in humans has been hampered by the difficulty of defining the BNM anatomically. Here, on the basis of a previous study that delineated the BNM of post-mortem human brains in a standard stereotaxic space, we sought to examine functional connectivity of the BNM, as compared to the nucleus accumbens (or ventral striatum, VS), in a large resting state functional magnetic resonance imaging data set. The BNM and VS shared but also showed a distinct pattern of cortical and subcortical connectivity. Compared to the VS, the BNM showed stronger positive connectivity with the putamen, pallidum, thalamus, amygdala and midbrain, as well as the anterior cingulate cortex, supplementary motor area and pre-supplementary motor area, a network of brain regions that respond to salient stimuli and orchestrate motor behavior. In contrast, compared to the BNM, the VS showed stronger positive connectivity with the ventral caudate and medial orbitofrontal cortex, areas implicated in reward processing and motivated behavior. Furthermore, the BNM and VS each showed extensive negative connectivity with visual and lateral prefrontal cortices. Together, the distinct cerebral functional connectivities support the role of the BNM in arousal, saliency responses and cognitive motor control and the VS in reward related behavior. Considering the importance of BNM in age-related cognitive decline, we explored the effects of age on BNM and VS connectivities. BNM connectivity to the visual and somatomotor cortices decreases while connectivity to subcortical structures including the midbrain, thalamus, and pallidum increases with age. These findings of age-related changes of cerebral functional connectivity of the BNM may facilitate research of the neural bases of cognitive decline in health and illness.

Details DOI

YNIMG Journal 2013 Journal Article

Task-related concurrent but opposite modulations of overlapping functional networks as revealed by spatial ICA

  • Jiansong Xu
  • Sheng Zhang
  • Vince D. Calhoun
  • John Monterosso
  • Chiang-shan R. Li
  • Patrick D. Worhunsky
  • Michael Stevens
  • Godfrey D. Pearlson

Animal studies indicate that different functional networks (FNs), each with a unique timecourse, may overlap at common brain regions. For understanding how different FNs overlap in the human brain and how the timecourses of overlapping FNs are modulated by cognitive tasks, we applied spatial independent component analysis (sICA) to functional magnetic resonance imaging (fMRI) data. These data were acquired from healthy participants while they performed a visual task with parametric loads of attention and working memory. sICA identified a total of 14 FNs, and they showed different extents of overlap at a majority of brain regions exhibiting any functional activity. More FNs overlapped at the higher-order association cortex including the anterior and posterior cingulate, precuneus, insula, and lateral and medial frontoparietal cortices (FPCs) than at the primary sensorimotor cortex. Furthermore, overlapping FNs exhibited concurrent but different task-related modulations of timecourses. FNs showing task-related up- vs. down-modulation of timecourses overlapped at both the lateral and medial FPCs and subcortical structures including the thalamus, striatum, and midbrain ventral tegmental area (VTA). Such task-related, concurrent, but opposite changes in timecourses in the same brain regions may not be detected by current analyses based on General-Linear-Model (GLM). The present findings indicate that multiple cognitive processes may associate with common brain regions and exhibit simultaneous but different modulations in timecourses during cognitive tasks.

Details DOI

YNIMG Journal 2012 Journal Article

Cerebral correlates of skin conductance responses in a cognitive task

  • Sheng Zhang
  • Sien Hu
  • Herta H. Chao
  • Xi Luo
  • Olivia M. Farr
  • Chiang-shan R. Li

Changes in physiological arousal frequently accompany cognitive performance. Many studies sought to identify the neural correlates of heightened arousal as indexed by skin conductance responses (SCR). However, the observed regional activations may be confounded by task events. We addressed this issue by recording SCR in 25 adults performing a stop signal task (SST) during functional magnetic resonance imaging. We compared only go trials with high and low SCR in order to isolate the event-independent processes. Furthermore, we distinguished go trials that followed another go, a stop success, or a stop error trial to examine whether the neural activities are contingent on the local context in which changes in SCR occurred. The results showed that the supplementary motor area responded to increased SCR irrespective of the preceding trial. The dorsal anterior cingulate cortex increased activation to heightened arousal most significantly in response to stop errors. The medial prefrontal cortex increased activation to SCR following a stop error but decreased activation following a go or stop success trial. These new findings specify the regional activations that accompany changes in physiological arousal during the SST and support distinct processes for the changes that occur under different local contexts. In particular, the MPFC shows opposing responses by increasing activation to changes in arousal evoked by salient stimuli and decreasing activation to the control of arousal.

Details DOI

YNIMG Journal 2012 Journal Article

Decreased saliency processing as a neural measure of Barratt impulsivity in healthy adults

  • Olivia M. Farr
  • Sien Hu
  • Sheng Zhang
  • Chiang-shan R. Li

Cognitive control is necessary to navigating through an uncertain world. With the stop signal task (SST), we measure how cognitive control functions in a controlled environment. There has been conflicting evidence on whether trait impulsivity might reflect differences in cognitive control during the SST. While some studies find that trait impulsivity relates to measures of response inhibition, such as the stop signal reaction time (SSRT), other studies do not. Here, in 92 young adult participants (58 females; age 25±4years), we examined whether trait impulsivity, measured by the Barratt impulsivity scale (BIS-11), is associated with differences in performance and regional brain activations for the component processes of cognitive control during the SST. Across participants, trait impulsivity showed a trend-level correlation with SSRT (F(1. 90)=3. 18, p<. 07; Pearson regression). In simple regressions, activation of the right anterior dorsal insula and middle frontal cortex (MFC) during stop as compared to go trials negatively correlated with motor and non-planning impulsivity score. Using the generalized form of psychophysiological interaction (gPPI), we showed that functional connectivity of the right insula and MFC with the left dorsolateral prefrontal cortex and bilateral visual areas were also negatively correlated with impulsivity. None of the other component processes of cognitive control, including response inhibition, error processing, post-error slowing, were significantly related to Barratt impulsivity. These results suggest that trait impulsivity as measured by BIS-11 may have distinct effects on saliency processing in adult individuals.

Details DOI

YNIMG Journal 2012 Journal Article

Functional connectivity mapping of the human precuneus by resting state fMRI

  • Sheng Zhang
  • Chiang-shan R. Li

Precuneus responds to a wide range of cognitive processes. Here, we examined how the patterns of resting state connectivity may define functional subregions in the precuneus. Using a K-means algorithm to cluster the whole-brain “correlograms” of the precuneus in 225 adult individuals, we corroborated the dorsal-anterior, dorsal-posterior, and ventral subregions, each involved in spatially guided behaviors, mental imagery, and episodic memory as well as self-related processing, with the ventral precuneus being part of the default mode network, as described extensively in earlier work. Furthermore, we showed that the lateral/medial volumes of dorsal anterior and dorsal posterior precuneus are each connected with areas of motor execution/attention and motor/visual imagery, respectively. Compared to the ventral precuneus, the dorsal precuneus showed greater connectivity with occipital and posterior parietal cortices, but less connectivity with the medial superior frontal and orbitofrontal gyri, anterior cingulate cortex as well as the parahippocampus. Compared to dorsal-posterior and ventral precuneus, the dorsal-anterior precuneus showed greater connectivity with the somatomotor cortex, as well as the insula, supramarginal, Heschl's, and superior temporal gyri, but less connectivity with the angular gyrus. Compared to ventral and dorsal-anterior precuneus, dorsal-posterior precuneus showed greater connectivity with the middle frontal gyrus. Notably, the precuneus as a whole has negative connectivity with the amygdala and the lateral and inferior orbital frontal gyri. Finally, men and women differed in the connectivity of precuneus. Men and women each showed greater connectivity with the dorsal precuneus in the cuneus and medial thalamus, respectively. Women also showed greater connectivity with ventral precuneus in the hippocampus/parahippocampus, middle/anterior cingulate gyrus, and middle occipital gyrus, compared to men. Taken together, these new findings may provide a useful platform upon which to further investigate sex-specific functional neuroanatomy of the precuneus and to elucidate the pathology of many neurological illnesses.

Details DOI

YNIMG Journal 2010 Journal Article

A neural measure of behavioral engagement: Task-residual low-frequency blood oxygenation level-dependent activity in the precuneus

  • Sheng Zhang
  • Chiang-shan Ray Li

Brain imaging has provided a useful tool to examine the neural processes underlying human cognition. A critical question is whether and how task engagement influences the observed regional brain activations. Here we highlighted this issue and derived a neural measure of task engagement from the task-residual low-frequency blood oxygenation level-dependent (BOLD) activity in the precuneus. Using independent component analysis, we identified brain regions in the default circuit – including the precuneus and medial prefrontal cortex (mPFC) – showing greater activation during resting as compared to task residuals in 33 individuals. Time series correlations with the posterior cingulate cortex as the seed region showed that connectivity with the precuneus was significantly stronger during resting as compared to task residuals. We hypothesized that if the task-residual BOLD activity in the precuneus reflects engagement, it should account for a certain amount of variance in task-related regional brain activation. In an additional experiment of 59 individuals performing a stop signal task, we observed that the fractional amplitude of low-frequency fluctuation (fALFF) of the precuneus but not the mPFC accounted for approximately 10% of the variance in prefrontal activation related to attentional monitoring and response inhibition. Taken together, these results suggest that task-residual fALFF in the precuneus may be a potential indicator of task engagement. This measurement may serve as a useful covariate in identifying motivation-independent neural processes that underlie the pathogenesis of a psychiatric or neurological condition.

Details DOI
v2026.09.13