Author name cluster

Yun Li

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.

25 papers

2 author rows

EAAI Journal 2026 Journal Article

Kolmogorov–Arnold network-based adaptive control allocation for overactuated systems with Lyapunov-stable learning

Jingxian Liao
Chenkai Cao
Yun Li

Adaptive control allocation is critical for overactuated systems to compensate for actuator faults without explicit fault detection logic. Traditional approaches such as pseudo-inverse and optimization-based methods provide fixed redistribution rules but cannot adapt to time-varying degradation. Recent neural-network allocators introduce online adaptivity through multi-layer perceptrons (MLPs), yet none provide closed-loop stability guarantees during learning. This paper introduces a Kolmogorov–Arnold Network (KAN)-based adaptive control allocation framework (allocKAN) with Lyapunov-stable online learning. We derive a composite Lyapunov stability framework showing that KAN parameter sensitivity scales only with network width and input dimension, independent of spline resolution, whereas MLP sensitivity grows with total parameter count. This directly determines the feasibility of the critical stability condition. Uniformly ultimately bounded stability theorems show that KAN yields tighter ultimate error bounds that scale only with architectural width, whereas MLP bounds scale with all learnable parameters. Offline experiments across 63 matched-budget configurations demonstrate 75. 2% lower validation error for allocKAN. Critically, increasing spline resolution improves approximation without sacrificing stability margins, a resolution-independence property unique to KANs. A 252-case closed-loop study on a ducted-fan unmanned aerial vehicle spanning 7 allocators, 3 controllers, and 12 fault scenarios confirms 100% bounded compliance for both neural allocators, while traditional methods diverge under severe faults. AllocKAN achieves 42%–77% lower parameter sensitivity, approximately 10 times tighter Lyapunov bounds, 37. 6–46. 6% lower final-state norms, and 50% faster fault recovery, while both allocators exceed the 400Hz real-time requirement with over 43% margin. These results establish KAN as a certifiable architecture for safety-critical adaptive allocation without controller redesign.

Details DOI

EAAI Journal 2026 Journal Article

SubAttack: A word-level adversarial textual attack method via antonym substitution

Chenqi Hua
Xiaojian Liu
Yi Zhu
Chaowei Zhang
Yun Li
Yunhao Yuan
Jipeng Qiang

Over the past few years, various word-level textual attack approaches have been proposed to reveal the vulnerability in existing deep neural networks and even large language models (LLMs) for Natural Language Processing (NLP). The textual attack aims to fool existing models into making erroneous predictions by altering the text without affecting the user’s understanding. However, current methods either struggle to construct semantically preserved adversarial texts and altered the semantics of the original text, or fail to consider the semantic perturbation constraints and are prone to invalid adversarial examples. In this paper, we propose an efficient and effective framework SubAttack to address these issues. SubAttack is a word-level adversarial textual attack method via antonym substitution, which replaces semantic indicator keywords to generate high-quality adversarial samples with considering both semantically preservation and semantic perturbation. Specifically, the process first involves tokenizing the text and performing part-of-speech tagging Identifying the semantic indicator keywords. Then, the antonym ranking is designed to decide the substitutions of candidate words to fit the context. Finally, while retaining the original text, the ranked antonyms are integrated into the text and the instructions are added for both semantically preservation and semantic perturbation. Extensive experiments reveal that state-of-the-art (SOTA) LLMs (e. g. Llama and QWen) are still vulnerable to our SubAttack. Further experiments show that the adversarial examples crafted by SubAttack usually have higher quality, exhibit better fluency and barely affect human performance and can bring more robustness improvement to victim models by adversarial training.

Details DOI

EAAI Journal 2026 Journal Article

Temporal point-supervised signal reconstruction: A human-annotation-free framework for weak moving target detection

Weihua Gao
Chunxu Ren
Jie Tang
Yun Li
Wenlong Niu
Xiaodong Peng

In low-altitude surveillance and early warning systems, detecting weak moving targets remains a significant challenge due to low signal energy, small spatial extent, and complex background clutter. Existing methods struggle with extracting robust features and suffer from the lack of reliable annotations. To address these limitations, we propose a novel Temporal Point-Supervised (TPS) framework that enables high-performance detection of weak targets without any manual annotations. Instead of conventional frame-based detection, our framework reformulates the task as a pixel-wise temporal signal modeling problem, where weak targets manifest as short-duration pulse-like responses. A Temporal Signal Reconstruction Network (TSRNet) is developed under the TPS paradigm to reconstruct these transient signals. TSRNet adopts an encoder–decoder architecture and integrates a dynamic multi-scale attention module to enhance its sensitivity to diverse temporal patterns. Additionally, a graph-based trajectory mining strategy is employed to suppress false alarms and ensure temporal consistency. Extensive experiments on a purpose-built low-signal-to-noise ratio dataset demonstrate that our framework outperforms state-of-the-art methods while requiring no human annotations. It achieves strong detection performance and operates at over 1000 frames per second, underscoring its potential for real-time deployment in practical scenarios.

Details DOI

EAAI Journal 2025 Journal Article

A domain adaptation method to Defend Chinese textual adversarial attacks via prompt-tuning

Yi Zhu
Zhenglong Li
Yun Li
Yunhao Yuan
Jipeng Qiang

The textual adversarial attack aims to fool existing models into making erroneous predictions by adding strategic perturbations to normal data without affecting the user’s understanding. Recently, methods based on Pre-trained Language Models (PLMs) and Large Language Models (LLMs) have shown promising performance in various Natural Language Processing (NLP) downstream tasks. However, due to significant deviations between the original and perturbed texts, these methods struggle to achieve satisfactory results in defending against textual adversarial attacks, especially in Chinese, which has unique syntactic structures. To address this issue, we propose a domain adaptation method for defending against Chinese textual adversarial attacks through a prompt-tuning model, which effectively mitigates the discrepancy between different domains. Specifically, the original and perturbed texts are treated as the source and target domains, respectively, with the textual adversarial defense task framed as a cross-domain classification problem. The soft prompt-tuning model trained in the source domain is iteratively adapted to uncover the true label information in the target domain. The graph attention network is incorporated to integrate Chinese syntactic structure information with semantic features. Through a voting mechanism on predicted labels generated by the iterative model, soft prompt-tuning is further optimized for cross-domain classification tasks. Extensive experimental results demonstrate the superior effectiveness of our method in Chinese textual adversarial defense tasks compared to baseline methods, including the state-of-the-art fine-tuning approaches for PLMs and LLMs.

Details DOI

EAAI Journal 2025 Journal Article

An efficient m-step lookahead rollout algorithm for profit-oriented selective disassembly sequence planning with operation stochastic failure

Yaping Ren
Leilei Meng
Guangdong Tian
Zhiwu Li
Yun Li

To fully reclaim large amounts of used electro-mechanic products, an effective and high-quality disassembly sequence is crucial that separates a product into parts/components one-by-one for recovery. However, it becomes rather challenging to efficiently determine the optimal/near-optimal disassembly sequence for used products since the quality conditions of used products are uncertain and disassembly operations that constitute the disassembly sequence might fail. In this paper, the stochastic failure characteristics of disassembly operations is taken into account and a selective disassembly sequence planning is studied for maximizing the recovering profit of used products. First, we formally model a profit-oriented selective disassembly sequence planning with operation stochastic failure (PSDSP-OSF), in which the expected recovering profit of a disassembly sequence is formulated based on the stochastic failure characteristics of disassembly operations. Then, we simplify and reformulate PSDSP-OSF according to the disassembly rule. A m-step lookahead rollout algorithm is proposed to efficiently solve PSDSP-OSF, in which every disassembly operation is selected by a lookahead decision rule that looking ahead from the current decision stage to future m decision stages. Finally, three different scales of products are selected as case studies and two well-known existing methods are employed to compare with the proposed approach. The computational results demonstrate that our proposed approach is significantly superior to the existing methods in solution quality, computational efficiency, and algorithm robustness.

Details DOI

AAAI Conference 2025 Conference Paper

Cross-Lingual Text-Rich Visual Comprehension: An Information Theory Perspective

Xinmiao Yu
Xiaocheng Feng
Yun Li
Minghui Liao
Ya-Qi Yu
Xiachong Feng
Weihong Zhong
Ruihan Chen

Recent Large Vision-Language Models (LVLMs) have shown promising reasoning capabilities on text-rich images from charts, tables, and documents. However, the abundant text within such images may increase the model's sensitivity to language. This raises the need to evaluate LVLM performance on cross-lingual text-rich visual inputs, where the language in the image differs from the language of the instructions. To address this, we introduce XT-VQA (Cross-Lingual Text-Rich Visual Question Answering), a benchmark designed to assess how LVLMs handle language inconsistency between image text and questions. XT-VQA integrates five existing text-rich VQA datasets and a newly collected dataset, XPaperQA, covering diverse scenarios that require faithful recognition and comprehension of visual information despite language inconsistency. Our evaluation of prominent LVLMs on XT-VQA reveals a significant drop in performance for cross-lingual scenarios, even for models with multilingual capabilities. A mutual information analysis suggests that this performance gap stems from cross-lingual questions failing to adequately activate relevant visual information. To mitigate this issue, we propose MVCL-MI (Maximization of Vision-Language Cross-Lingual Mutual Information), where a visual-text cross-lingual alignment is built by maximizing mutual information between the model's outputs and visual information. This is achieved by distilling knowledge from monolingual to cross-lingual settings through KL divergence minimization, where monolingual output logits serve as a teacher. Experimental results on the XT-VQA demonstrate that MVCL-MI effectively reduces the visual-text cross-lingual performance disparity while preserving the inherent capabilities of LVLMs, shedding new light on the potential practice for improving LVLMs.

PDF Details DOI

EAAI Journal 2025 Journal Article

Design of an efficient fault-tolerant quantum-computing circuit with quantum neural network learning

Chen Lin
Rucong Xu
Yun Li

Fault-tolerance is key to the practical realization of quantum computation, but the design of an efficient, low-overhead and fault-tolerant error-correction circuit remains a major challenge so far. To help address this issue, we first propose a noise-adaptive dissipative quantum neural network (DQNN) model to mitigate the effects of error propagation for constructing a fault-tolerant quantum circuit. Then, we develop a method for preparing a fault-tolerant auxiliary entangled state based on the DQNN model, reducing computational delays and qubit resource consumption. This method utilizes the adaptability of quantum machine learning to the distribution of noisy inputs and its practicality in noisy intermediate scale quantum devices, thus reducing interaction with classical computers and further optimizing the real-time requirements for active error-correction. By integrating quantum error-correction and quantum neural network learning, this DQNN scheme provides a novel solution for constructing scalable fault-tolerant quantum computation. Compared with existing fault-tolerant methods, the DQNN process requires fewer error-propagation efforts and offers higher fidelity in a noisy environment for error thresholds higher than 1 0 − 4. The effectiveness of this method is verified through experimental simulations using the Qiskit. The code for experiments and model in this paper can be found on GitHub: https: //github. com/Ricardo-Vv/Qiskit_exam/tree/master.

Details DOI

AAAI Conference 2025 Conference Paper

Is LLMs Hallucination Usable? LLM-based Negative Reasoning for Fake News Detection

Chaowei Zhang
Zongling Feng
Zewei Zhang
Jipeng Qiang
Guandong Xu
Yun Li

The questionable responses caused by knowledge hallucination may lead to LLMs' unstable ability in decision-making. However, it has never been investigated whether the LLMs' hallucination is possibly usable for generating negative reasoning to assist fake news detection. In this paper, we propose a novel supervised self-reinforced reasoning rectification approach - SR^3 that not only yields common reasonable reasoning for news but also forces LLMs to generate the wrong understandings of news via LLMs reflection for semantic consistency learning. Upon that, we construct a negative reasoning-based news learning model called - NRFE, which leverages positive or negative news-reasoning pairs for learning the semantic consistency between them. To avoid the impact of label-implicated reasoning, we deploy a student model - NRFE-D that only takes news content as input to inspect the performance of our method by distilling the knowledge from NRFE. The experimental results verified on three popular fake news datasets demonstrate the superiority of our method compared with three kinds of baselines including prompting-based LLMs, fine-tuning-based PLMs, and other representative fake news detection methods.

PDF Details DOI

IROS Conference 2025 Conference Paper

Multi-PrefDrive: Optimizing Large Language Models for Autonomous Driving Through Multi-Preference Tuning

Yun Li
Ehsan Javanmardi
Simon Thompson
Kai Katsumata
Alex Orsholits
Manabu Tsukada

This paper introduces Multi-PrefDrive, a framework that significantly enhances LLM-based autonomous driving through multidimensional preference tuning. Aligning LLMs with human driving preferences is crucial yet challenging, as driving scenarios involve complex decisions where multiple incorrect actions can correspond to a single correct choice. Traditional binary preference tuning fails to capture this complexity. Our approach pairs each chosen action with multiple rejected alternatives, better reflecting real-world driving decisions. By implementing the Plackett-Luce preference model, we enable nuanced ranking of actions across the spectrum of possible errors. Experiments in the CARLA simulator demonstrate that our algorithm achieves an 11. 0% improvement in overall score and an 83. 6% reduction in infrastructure collisions, while showing perfect compliance with traffic signals in certain environments. Comparative analysis against DPO and its variants reveals that Multi-PrefDrive’s superior discrimination between chosen and rejected actions, which achieving a margin value of 25, and such ability has been directly translates to enhanced driving performance. We implement memory-efficient techniques including LoRA and 4-bit quantization to enable deployment on consumer-grade hardware and will open-source our training code and multi-rejected dataset to advance research in LLM-based autonomous driving systems. Project Page (https://liyun0607.github.io/).

Details

ICLR Conference 2025 Conference Paper

PEARL: Parallel Speculative Decoding with Adaptive Draft Length

Tianyu Liu
Yun Li
Qitan Lv
Kai Liu 0052
Jianchen Zhu
Winston Hu
Xiao Sun

Speculative decoding (SD), where an extra draft model is employed to provide multiple **draft** tokens first and then the original target model verifies these tokens in parallel, has shown great power for LLM inference acceleration. However, existing SD methods suffer from the mutual waiting problem, i.e., the target model gets stuck when the draft model is *guessing* tokens, and vice versa. This problem is directly incurred by the asynchronous execution of the draft model and the target model, and is exacerbated due to the fixed draft length in speculative decoding. To address these challenges, we propose a conceptually simple, flexible, and general framework to boost speculative decoding, namely **P**arallel sp**E**culative decoding with **A**daptive d**R**aft **L**ength (PEARL). Specifically, PEARL proposes *pre-verify* to verify the first draft token in advance during the drafting phase, and *post-verify* to generate more draft tokens during the verification phase. PEARL parallels the drafting phase and the verification phase via applying the two strategies, and achieves adaptive draft length for different scenarios, which effectively alleviates the mutual waiting problem. Experiments on various text generation benchmarks demonstrate the effectiveness of our PEARL, leading to a superior speedup performance up to **4.43$\times$** and **1.50$\times$**, compared to auto-regressive decoding and vanilla speculative decoding, respectively.

Details

EAAI Journal 2025 Journal Article

Photovoltaic system modeling and forecasting techniques: A survey

Chen Luo
Naji Al-Messabi
Zhaoqi Kuang
Changjiang Ma
Ibrahim El-Amin
Hui Deng
Yun Li

Modeling provides engineering science a vital technique in the third paradigm of science. This paper reviews a series of modeling techniques for forecasting solar energy yields of photovoltaic (PV) systems, with comparisons among various aspects of solar photovoltaic forecasting, including forecasting techniques, irradiance models, and PV forecasting software and services. Efficient integration of renewable energy sources, in particular PV systems, to the power gird complement fossil fueled generation in reducing emissions. A pivotal step in this integration is to predict the PV outputs, so as to account for their dynamic energy contribution in generation planning and unit commitment programs. Existing research has mostly been on forecasting solar irradiance figures from a very short term (minutes/hours) to a short term (days/weeks) horizon. The paper highlights the use of advanced intelligent modeling techniques. The results serve as a practical guide to renewable energy researchers and engineers for best available approaches in handling photovoltaic forecasting and renewable power generation.

Details DOI

EAAI Journal 2025 Journal Article

Soft Prompt-tuning with Self-Resource Verbalizer for short text streams

Yi Zhu
Ye Wang
Yun Li
Jipeng Qiang
Yunhao Yuan

Short text streams such as real-time news and search snippets have attained vast amounts of attention and research in recent decades, the characteristics of high generation velocity, feature sparsity, and high ambiguity accentuate both the importance and challenges to language models. However, most of the existing short text stream classification methods can neither automatically select relevant knowledge components for arbitrary samples, nor expand knowledge internally instead of rely on external open knowledge base to address the inherent limitations of short text stream. In this paper, we propose a Soft Prompt-tuning with Self-Resource Verbalizer (SPSV for short) for short text stream classification, the soft prompt with self-resource knowledgeable expansion is conducted for updating label words space to address evolved semantic topics in the data streams. Specifically, the automatic constructed prompt is first generated to instruct the model prediction, which is optimized to address the problem of high velocity and topic drift in short text streams. Then, in each chunk, the projection between category names and label words space, i. e. verbalizer, is updated, which is constructed by internal knowledge expansion from the short text itself. Through comprehensive experiments on four well-known benchmark datasets, we validate the superb performance of our method compared to other short text stream classification and fine-tuning PLMs methods, which achieves up to more than 90% classification accuracy with the counts of data chunk increased.

Details DOI

NeurIPS Conference 2024 Conference Paper

Calibrated Self-Rewarding Vision Language Models

Yiyang Zhou
Zhiyuan Fan
Dongjie Cheng
Sihan Yang
Zhaorun Chen
Chenhang Cui
Xiyao Wang
Yun Li

Large Vision-Language Models (LVLMs) have made substantial progress by integrating pre-trained large language models (LLMs) and vision models through instruction tuning. Despite these advancements, LVLMs often exhibit the hallucination phenomenon, where generated text responses appear linguistically plausible but contradict the input image, indicating a misalignment between image and text pairs. This misalignment arises because the model tends to prioritize textual information over visual input, even when both the language model and visual representations are of high quality. Existing methods leverage additional models or human annotations to curate preference data and enhance modality alignment through preference optimization. These approaches are resource-intensive and may not effectively reflect the target LVLM's preferences, making the curated preferences easily distinguishable. Our work addresses these challenges by proposing the Calibrated Self-Rewarding (CSR) approach, which enables the model to self-improve by iteratively generating candidate responses, evaluating the reward for each response, and curating preference data for fine-tuning. In the reward modeling, we employ a step-wise strategy and incorporate visual constraints into the self-rewarding process to place greater emphasis on visual input. Empirical results demonstrate that CSR significantly enhances performance and reduces hallucinations across twelve benchmarks and tasks, achieving substantial improvements over existing methods by 7. 62\%. Our empirical results are further supported by rigorous theoretical analysis, under mild assumptions, verifying the effectiveness of introducing visual constraints into the self-rewarding paradigm. Additionally, CSR shows compatibility with different vision-language models and the ability to incrementally improve performance through iterative fine-tuning.

PDF Details DOI

NeurIPS Conference 2024 Conference Paper

CARES: A Comprehensive Benchmark of Trustworthiness in Medical Vision Language Models

Peng Xia
Ze Chen
Juanxi Tian
Yangrui Gong
Ruibo Hou
Yue Xu
Zhenbang Wu
Zhiyuan Fan

Artificial intelligence has significantly impacted medical applications, particularly with the advent of Medical Large Vision Language Models (Med-LVLMs), sparking optimism for the future of automated and personalized healthcare. However, the trustworthiness of Med-LVLMs remains unverified, posing significant risks for future model deployment. In this paper, we introduce CARES and aim to comprehensively evaluate the Trustworthiness of Med-LVLMs across the medical domain. We assess the trustworthiness of Med-LVLMs across five dimensions, including trustfulness, fairness, safety, privacy, and robustness. CARES comprises about 41K question-answer pairs in both closed and open-ended formats, covering 16 medical image modalities and 27 anatomical regions. Our analysis reveals that the models consistently exhibit concerns regarding trustworthiness, often displaying factual inaccuracies and failing to maintain fairness across different demographic groups. Furthermore, they are vulnerable to attacks and demonstrate a lack of privacy awareness. We publicly release our benchmark and code in https: //github. com/richard-peng-xia/CARES.

PDF Details DOI

NeurIPS Conference 2024 Conference Paper

STimage-1K4M: A histopathology image-gene expression dataset for spatial transcriptomics

Jiawen Chen
Muqing Zhou
Wenrong Wu
Jinwei Zhang
Yun Li
Didong Li

Recent advances in multi-modal algorithms have driven and been driven by the increasing availability of large image-text datasets, leading to significant strides in various fields, including computational pathology. However, in most existing medical image-text datasets, the text typically provides high-level summaries that may not sufficiently describe sub-tile regions within a large pathology image. For example, an image might cover an extensive tissue area containing cancerous and healthy regions, but the accompanying text might only specify that this image is a cancer slide, lacking the nuanced details needed for in-depth analysis. In this study, we introduce STimage-1K4M, a novel dataset designed to bridge this gap by providing genomic features for sub-tile images. STimage-1K4M contains 1, 149 images derived from spatial transcriptomics data, which captures gene expression information at the level of individual spatial spots within a pathology image. Specifically, each image in the dataset is broken down into smaller sub-image tiles, with each tile paired with $15, 000-30, 000$ dimensional gene expressions. With $4, 293, 195$ pairs of sub-tile images and gene expressions, STimage-1K4M offers unprecedented granularity, paving the way for a wide range of advanced research in multi-modal data analysis an innovative applications in computational pathology, and beyond.

PDF Details DOI

TIST Journal 2023 Journal Article

Cost-sensitive Tensor-based Dual-stage Attention LSTM with Feature Selection for Data Center Server Power Forecasting

Ziyu Shen
Binghui Liu
Qing Zhou
Zheng Liu
Bin Xia
Yun Li

Power forecasting has a guiding effect on power-aware scheduling strategies to reduce unnecessary power consumption in data centers. Many metrics related to power consumption can be collected in physical servers, such as the status of CPU, memory, and other components. However, most existing methods empirically exploit a small number of metrics to forecast power consumption. To this end, this article uses feature selection based on causality to explore the metrics that strongly influence the power consumption of different tasks. Moreover, we propose a tensor-based dual-stage attention LSTM to forecast the non-linear and non-periodic power consumption. In the proposed model, a multi-way delay embedding transform is utilized to convert the time series into tensors along the temporal direction. The LSTM combines with the tensor technique and the attention mechanism to capture the temporal pattern effectively. In addition, we adopt the cost-sensitive loss function to optimize the specific power forecasting problem in data centers. The experimental results demonstrate that our method can achieve up to 1.4% to 4.3% forecasting accuracy improvement compared with the state-of-the-art models.

Details DOI

JBHI Journal 2023 Journal Article

Disentangled and Side-Aware Unsupervised Domain Adaptation for Cross-Dataset Subjective Tinnitus Diagnosis

Yun Li
Zhe Liu
Lina Yao
Jessica J. M. Monaghan
David McAlpine

EEG-based tinnitus classification is a valuable tool for tinnitus diagnosis, research, and treatments. Most current works are limited to a single dataset where data patterns are similar. But EEG signals are highly non-stationary, resulting in model's poor generalization to new users, sessions or datasets. Thus, designing a model that can generalize to new datasets is beneficial and indispensable. To mitigate distribution discrepancy across datasets, we propose to achieve Disentangled and Side-aware Unsupervised Domain Adaptation (DSUDA) for cross-dataset tinnitus diagnosis. A disentangled auto-encoder is developed to decouple class-irrelevant information from the EEG signals to improve the classifying ability. The side-aware unsupervised domain adaptation module adapts the class-irrelevant information as domain variance to a new dataset and excludes the variance to obtain the class-distill features for the new dataset classification. It also aligns signals of left and right ears to overcome inherent EEG pattern difference. We compare DSUDA with state-of-the-art methods, and our model achieves significant improvements over competitors regarding comprehensive evaluation criteria. The results demonstrate our model can successfully generalize to a new dataset and effectively diagnose tinnitus.

Details DOI

AIJ Journal 2023 Journal Article

Natural language watermarking via paraphraser-based lexical substitution

Jipeng Qiang
Shiyu Zhu
Yun Li
Yi Zhu
Yunhao Yuan
Xindong Wu

Although powerful pretrained language models generate high-quality output text, they bring new concerns about the potential misuse of such models for malicious purposes. Natural language watermarking (NLW) is a technique that is desgined to help tracing the provenance of texts for againsting possible attacks, where the watermark signals are embedded into cover texts using synonym substitutions. The up-to-date BERT-based NLW methods have made remarkable progress on performance improvement of watermarking through generating substitutes for a masked target word. Yet, the BERT-based NLWs focus on the context of texts rather than the meaning of target words, which might make the capacity of watermark embeddings being lower. To address the limitations, this study proposes a novel NLW method by incorporating a paraphraser-based lexical substitution method. Under the promise of paraphrase preservation, the proposed NLW method utilizes the knowledge of paraphrase modeling to generate the substitute candidates to replace the words in original sentences capable of carrying the watermark signal in local contexts. We empirically show that our NLW method not only has a better meaning-preserved, but improves the payload more than 2 times compared with the BERT-based NLW method. Besides, compared with previous state-of-the-art method Compared with other state-of-the-art baselines, the experimental results show that the proposed LS method improves the Precision@1 score from 51. 7% to 58. 3% and from 50. 5% to 62. 6% on LS07 and CoInCo benchmarks, respectively.

Details DOI

NeurIPS Conference 2023 Conference Paper

On the Identifiability and Interpretability of Gaussian Process Models

Jiawen Chen
Wancen Mu
Yun Li
Didong Li

In this paper, we critically examine the prevalent practice of using additive mixtures of Mat\'ern kernels in single-output Gaussian process (GP) models and explore the properties of multiplicative mixtures of Mat\'ern kernels for multi-output GP models. For the single-output case, we derive a series of theoretical results showing that the smoothness of a mixture of Mat\'ern kernels is determined by the least smooth component and that a GP with such a kernel is effectively equivalent to the least smooth kernel component. Furthermore, we demonstrate that none of the mixing weights or parameters within individual kernel components are identifiable. We then turn our attention to multi-output GP models and analyze the identifiability of the covariance matrix $A$ in the multiplicative kernel $K(x, y) = AK_0(x, y)$, where $K_0$ is a standard single output kernel such as Mat\'ern. We show that $A$ is identifiable up to a multiplicative constant, suggesting that multiplicative mixtures are well suited for multi-output tasks. Our findings are supported by extensive simulations and real applications for both single- and multi-output settings. This work provides insight into kernel selection and interpretation for GP models, emphasizing the importance of choosing appropriate kernel structures for different tasks.

PDF Details

AAAI Conference 2021 Conference Paper

Task Aligned Generative Meta-learning for Zero-shot Learning

Zhe Liu
Yun Li
Lina Yao
Xianzhi Wang
Guodong Long

Zero-shot learning (ZSL) refers to the problem of learning to classify instances from novel classes (unseen) that are absent in the training set (seen). Most ZSL methods infer the correlation between visual features and attributes to train the classifier for unseen classes. They may have a strong bias towards seen classes during training. Meta-learning has been introduced to mitigate the basis, but meta-ZSL methods are inapplicable when tasks used for training are sampled from diverse distributions. In this regard, we propose a novel Task-aligned Generative Meta-learning model for Zeroshot learning (TGMZ), aiming to mitigate the potentially biased training and to enable meta-ZSL to accommodate realworld datasets that contain diverse distributions. Specifically, TGMZ incorporates an attribute-conditioned task-wise distribution alignment network that projects tasks into a unified distribution to deliver an unbiased model. Our experiments show TGMZ achieves a relative improvement of 2. 1%, 3. 0%, 2. 5%, and 7. 6% over state-of-the-art algorithms on AWA1, AWA2, CUB, and aPY datasets, respectively. Overall, TGMZ outperforms competitors by 3. 6% in the generalized zero-shot learning (GZSL) setting and 7. 9% in our proposed fusion- ZSL setting.

PDF Details

IJCAI Conference 2020 Conference Paper

CooBa: Cross-project Bug Localization via Adversarial Transfer Learning

Ziye Zhu
Yun Li
Hanghang Tong
Yu Wang

Bug localization plays an important role in software quality control. Many supervised machine learning models have been developed based on historical bug-fix information. Despite being successful, these methods often require sufficient historical data (i. e. , labels), which is not always available especially for newly developed software projects. In response, cross-project bug localization techniques have recently emerged whose key idea is to transferring knowledge from label-rich source project to locate bugs in the target project. However, a major limitation of these existing techniques lies in that they fail to capture the specificity of each individual project, and are thus prone to negative transfer. To address this issue, we propose an adversarial transfer learning bug localization approach, focusing on only transferring the common characteristics (i. e. , public information) across projects. Specifically, our approach (CooBa) learns the indicative public information from cross-project bug reports through a shared encoder, and extracts the private information from code files by an individual feature extractor for each project. CooBa further incorporates adversarial learning mechanism to ensure that public information shared between multiple projects could be effectively extracted. Extensive experiments on four large-scale real-world data sets demonstrate that the proposed CooBa significantly outperforms the state of the art techniques.

PDF Details DOI

AAAI Conference 2020 Conference Paper

Lexical Simplification with Pretrained Encoders

Jipeng Qiang
Yun Li
Yi Zhu
Yunhao Yuan
Xindong Wu

Lexical simpliﬁcation (LS) aims to replace complex words in a given sentence with their simpler alternatives of equivalent meaning. Recently unsupervised lexical simpliﬁcation approaches only rely on the complex word itself regardless of the given sentence to generate candidate substitutions, which will inevitably produce a large number of spurious candidates. We present a simple LS approach that makes use of the Bidirectional Encoder Representations from Transformers (BERT) which can consider both the given sentence and the complex word during generating candidate substitutions for the complex word. Speciﬁcally, we mask the complex word of the original sentence for feeding into the BERT to predict the masked token. The predicted results will be used as candidate substitutions. Despite being entirely unsupervised, experimental results show that our approach obtains obvious improvement compared with these baselines leveraging linguistic databases and parallel corpus, outperforming the state-of-the-art by more than 12 Accuracy points on three well-known benchmarks.

PDF Details

YNICL Journal 2020 Journal Article

Reorganization of rich-clubs in functional brain networks during propofol-induced unconsciousness and natural sleep

Shengpei Wang
Yun Li
Shuang Qiu
Chuncheng Zhang
Guyan Wang
Junfang Xian
Tianzuo Li
Huiguang He

BACKGROUND: General anesthesia (GA) provides an invaluable experimental tool to understand the essential neural mechanisms underlying consciousness. Previous neuroimaging studies have shown the functional integration and segregation of brain functional networks during anesthetic-induced alteration of consciousness. However, the organization pattern of hubs in functional brain networks remains unclear. Moreover, comparisons with the well-characterized physiological unconsciousness can help us understand the neural mechanisms of anesthetic-induced unconsciousness. METHODS: Resting-state functional magnetic resonance imaging was performed during wakefulness, mild propofol-induced sedation (m-PIS), and deep PIS (d-PIS) with clinical unconsciousness on 8 healthy volunteers and wakefulness and natural sleep on 9 age- and sex-matched healthy volunteers. Large-scale functional brain networks of each volunteer were constructed based on 160 regions of interest. Then, rich-club organizations in brain functional networks and nodal properties (nodal strength and efficiency) were assessed and analyzed among the different states and groups. RESULTS: Rich-clubs in the functional brain networks were reorganized during alteration of consciousness induced by propofol. Firstly, rich-club nodes were switched from the posterior cingulate cortex (PCC), angular gyrus, and anterior and middle insula to the inferior parietal lobule (IPL), inferior parietal sulcus (IPS), and cerebellum. When sedation was deepened to unconsciousness, the rich-club nodes were switched to the occipital and angular gyrus. These results suggest that the rich-club nodes were switched among the high-order cognitive function networks (default mode network [DMN] and fronto-parietal network [FPN]), sensory networks (occipital network [ON]), and cerebellum network (CN) from consciousness (wakefulness) to propofol-induced unconsciousness. At the same time, compared with wakefulness, local connections were switched to rich-club connections during propofol-induced unconsciousness, suggesting a strengthening of the overall information commutation of networks. Nodal efficiency of the anterior and middle insula and ventral frontal cortex was significantly decreased. Additionally, from wakefulness to natural sleep, a similar pattern of rich-club reorganization with propofol-induced unconsciousness was observed: rich-club nodes were switched from the DMN (including precuneus and PCC) to the sensorimotor network (SMN, including part of the frontal and temporal gyrus). Compared with natural sleep, nodal efficiency of the insula, frontal gyrus, PCC, and cerebellum significantly decreased during propofol-induced unconsciousness. CONCLUSIONS: Our study demonstrated that the rich-club reorganization in functional brain networks is characterized by switching of rich-club nodes between the high-order cognitive and sensory and motor networks during propofol-induced alteration of consciousness and natural sleep. These findings will help understand the common neurological mechanism of pharmacological and physiological unconsciousness.

Details DOI

YNIMG Journal 2014 Journal Article

Prestimulus alpha power predicts fidelity of sensory encoding in perceptual decision making

Bin Lou
Yun Li
Marios G. Philiastides
Paul Sajda

Pre-stimulus α power has been shown to correlate with the behavioral accuracy of perceptual decisions. In most cases, these correlations have been observed by comparing α power for different behavioral outcomes (e. g. correct vs incorrect trials). In this paper we investigate such covariation within the context of behaviorally-latent fluctuations in task-relevant post-stimulus neural activity. Specially we consider variations of pre-stimulus α power with post-stimulus EEG components in a two alternative forced choice visual discrimination task. EEG components, discriminative of stimulus class, are identified using a linear multivariate classifier and only the variability of the components for correct trials (regardless of stimulus class, and for nominally identical stimuli) are correlated with the corresponding pre-stimulus α power. We find a significant relationship between the mean and variance of the pre-stimulus α power and the variation of the trial-to-trial magnitude of an early post-stimulus EEG component. This relationship is not seen for a later EEG component that is also discriminative of stimulus class and which has been previously linked to the quality of evidence driving the decision process. Our results suggest that early perceptual representations, rather than temporally later neural correlates of the perceptual decision, are modulated by pre-stimulus state.

Details DOI

AAAI Conference 2012 Conference Paper

Ensemble Feature Weighting Based on Local Learning and Diversity

Yun Li
Suyan Gao
Songcan Chen

Recently, besides the performance, the stability (robustness, i. e. , the variation in feature selection results due to small changes in the data set) of feature selection is received more attention. Ensemble feature selection where multiple feature selection outputs are combined to yield more robust results without sacrificing the performance is an effective method for stable feature selection. In order to make further improvements of the performance (classification accuracy), the diversity regularized ensemble feature weighting framework is presented, in which the base feature selector is based on local learning with logistic loss for its robustness to huge irrelevant features and small samples. At the same time, the sample complexity of the proposed ensemble feature weighting algorithm is analyzed based on the VCtheory. The experiments on different kinds of data sets show that the proposed ensemble method can achieve higher accuracy than other ensemble ones and other stable feature selection strategy (such as sample weighting) without sacrificing stability.

PDF Details