Author name cluster

Yan Feng

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.

13 papers

1 author row

AAAI Conference 2025 Conference Paper

Advancing Loss Functions in Recommender Systems: A Comparative Study with a Rényi Divergence-Based Solution

Shengjia Zhang
Jiawei Chen
Changdong Li
Sheng Zhou
Qihao Shi
Yan Feng
Chun Chen
Can Wang

Loss functions play a pivotal role in optimizing recommendation models. Among various loss functions, Softmax Loss (SL) and Cosine Contrastive Loss (CCL) are particularly effective. Their theoretical connections and differences warrant in-depth exploration. This work conducts comprehensive analyses of these losses, yielding significant insights: 1) Common strengths --- both can be viewed as augmentations of traditional losses with Distributional Robust Optimization (DRO), enhancing robustness to distributional shifts; 2) Respective limitations --- stemming from their use of different distribution distance metrics in DRO optimization, SL exhibits high sensitivity to false negative instances, whereas CCL suffers from low data utilization. To address these limitations, this work proposes a new loss function, DrRL, which generalizes SL and CCL by leveraging Rényi-divergence in DRO optimization. DrRL incorporates the advantageous structures of both SL and CCL, and can be demonstrated to effectively mitigate their limitations. Extensive experiments have been conducted to validate the superiority of DrRL on both recommendation accuracy and robustness.

PDF Details DOI

AAAI Conference 2025 Conference Paper

Efficient Self-Supervised Video Hashing with Selective State Spaces

Jinpeng Wang
Niu Lian
Jun Li
Yuting Wang
Yan Feng
Bin Chen
Yongbing Zhang
Shu-Tao Xia

Self-supervised video hashing (SSVH) is a practical task in video indexing and retrieval. Although Transformers are predominant in SSVH for their impressive temporal modeling capabilities, they often suffer from computational and memory inefficiencies. Drawing inspiration from Mamba, an advanced state-space model, we explore its potential in SSVH to achieve a better balance between efficacy and efficiency. We introduce S5VH, a Mamba-based video hashing model with an improved self-supervised learning paradigm. Specifically, we design bidirectional Mamba layers for both the encoder and decoder, which are effective and efficient in capturing temporal relationships thanks to the data-dependent selective scanning mechanism with linear complexity. In our learning strategy, we transform global semantics in the feature space into semantically consistent and discriminative hash centers, followed by a center alignment loss as a global learning signal. Our self-local-global (SLG) paradigm significantly improves learning efficiency, leading to faster and better convergence. Extensive experiments demonstrate S5VH's improvements over state-of-the-art methods, superior transferability, and scalable advantages in inference efficiency.

PDF Details DOI

YNIMG Journal 2024 Journal Article

Microstate-based brain network dynamics distinguishing temporal lobe epilepsy patients: A machine learning approach

Zihan Wei
Xinpei Wang
Chao Liu
Yan Feng
Yajing Gan
Yuqing Shi
Xiaoli Wang
Yonghong Liu

Temporal lobe epilepsy (TLE) stands as the predominant adult focal epilepsy syndrome, characterized by dysfunctional intrinsic brain dynamics. However, the precise mechanisms underlying seizures in these patients remain elusive. Our study encompassed 116 TLE patients compared with 51 healthy controls. Employing microstate analysis, we assessed brain dynamic disparities between TLE patients and healthy controls, as well as between drug-resistant epilepsy (DRE) and drug-sensitive epilepsy (DSE) patients. We constructed dynamic functional connectivity networks based on microstates and quantified their spatial and temporal variability. Utilizing these brain network features, we developed machine learning models to discriminate between TLE patients and healthy controls, and between DRE and DSE patients. Temporal dynamics in TLE patients exhibited significant acceleration compared to healthy controls, along with heightened synchronization and instability in brain networks. Moreover, DRE patients displayed notably lower spatial variability in certain parts of microstate B, E and F dynamic functional connectivity networks, while temporal variability in certain parts of microstate E and G dynamic functional connectivity networks was markedly higher in DRE patients compared to DSE patients. The machine learning model based on these spatiotemporal metrics effectively differentiated TLE patients from healthy controls and discerned DRE from DSE patients. The accelerated microstate dynamics and disrupted microstate sequences observed in TLE patients mirror highly unstable intrinsic brain dynamics, potentially underlying abnormal discharges. Additionally, the presence of highly synchronized and unstable activities in brain networks of DRE patients signifies the establishment of stable epileptogenic networks, contributing to the poor responsiveness to antiseizure medications. The model based on spatiotemporal metrics demonstrated robust predictive performance, accurately distinguishing both TLE patients from healthy controls and DRE patients from DSE patients.

Details DOI

NeurIPS Conference 2024 Conference Paper

PSL: Rethinking and Improving Softmax Loss from Pairwise Perspective for Recommendation

Weiqin Yang
Jiawei Chen
Xin Xin
Sheng Zhou
Binbin Hu
Yan Feng
Chun Chen
Can Wang

Softmax Loss (SL) is widely applied in recommender systems (RS) and has demonstrated effectiveness. This work analyzes SL from a pairwise perspective, revealing two significant limitations: 1) the relationship between SL and conventional ranking metrics like DCG is not sufficiently tight; 2) SL is highly sensitive to false negative instances. Our analysis indicates that these limitations are primarily due to the use of the exponential function. To address these issues, this work extends SL to a new family of loss functions, termed Pairwise Softmax Loss (PSL), which replaces the exponential function in SL with other appropriate activation functions. While the revision is minimal, we highlight three merits of PSL: 1) it serves as a tighter surrogate for DCG with suitable activation functions; 2) it better balances data contributions; and 3) it acts as a specific BPR loss enhanced by Distributionally Robust Optimization (DRO). We further validate the effectiveness and robustness of PSL through empirical experiments. The code is available at https: //github. com/Tiny-Snow/IR-Benchmark.

PDF Details DOI

NeurIPS Conference 2023 Conference Paper

OpenGSL: A Comprehensive Benchmark for Graph Structure Learning

Zhiyao Zhou
Sheng Zhou
Bochao Mao
Xuanyi Zhou
Jiawei Chen
Qiaoyu Tan
Daochen Zha
Yan Feng

Graph Neural Networks (GNNs) have emerged as the de facto standard for representation learning on graphs, owing to their ability to effectively integrate graph topology and node attributes. However, the inherent suboptimal nature of node connections, resulting from the complex and contingent formation process of graphs, presents significant challenges in modeling them effectively. To tackle this issue, Graph Structure Learning (GSL), a family of data-centric learning approaches, has garnered substantial attention in recent years. The core concept behind GSL is to jointly optimize the graph structure and the corresponding GNN models. Despite the proposal of numerous GSL methods, the progress in this field remains unclear due to inconsistent experimental protocols, including variations in datasets, data processing techniques, and splitting strategies. In this paper, we introduce OpenGSL, the first comprehensive benchmark for GSL, aimed at addressing this gap. OpenGSL enables a fair comparison among state-of-the-art GSL methods by evaluating them across various popular datasets using uniform data processing and splitting strategies. Through extensive experiments, we observe that existing GSL methods do not consistently outperform vanilla GNN counterparts. We also find that there is no significant correlation between the homophily of the learned structure and task performance, challenging the common belief. Moreover, we observe that the learned graph structure demonstrates a strong generalization ability across different GNN models, despite the high computational and space consumption. We hope that our open-sourced library will facilitate rapid and equitable evaluation and inspire further innovative research in this field. The code of the benchmark can be found in https: //github. com/OpenGSL/OpenGSL.

PDF Details

AAAI Conference 2023 Conference Paper

Robust Sequence Networked Submodular Maximization

Qihao Shi
Bingyang Fu
Can Wang
Jiawei Chen
Sheng Zhou
Yan Feng
Chun Chen

In this paper, we study the Robust optimization for sequence Networked submodular maximization (RoseNets) problem. We interweave the robust optimization with the sequence networked submodular maximization. The elements are connected by a directed acyclic graph and the objective function is not submodular on the elements but on the edges in the graph. Under such networked submodular scenario, the impact of removing an element from a sequence depends both on its position in the sequence and in the network. This makes the existing robust algorithms inapplicable and calls for new robust algorithms. In this paper, we take the first step to study the RoseNets problem. We design a robust greedy algorithms, which is robust against the removal of an arbitrary subset of the selected elements. The approximation ratio of the algorithm depends both on the number of the removed elements and the network topology. We further conduct experiments on real applications of recommendation and link prediction. The experimental results demonstrate the effectiveness of the proposed algorithm.

PDF Details DOI

AAAI Conference 2021 Conference Paper

Cross-Layer Distillation with Semantic Calibration

Defang Chen
Jian-Ping Mei
Yuan Zhang
Can Wang
Zhe Wang
Yan Feng
Chun Chen

Recently proposed knowledge distillation approaches based on feature-map transfer validate that intermediate layers of a teacher model can serve as effective targets for training a student model to obtain better generalization ability. Existing studies mainly focus on particular representation forms for knowledge transfer between manually specified pairs of teacher-student intermediate layers. However, semantics of intermediate layers may vary in different networks and manual association of layers might lead to negative regularization caused by semantic mismatch between certain teacherstudent layer pairs. To address this problem, we propose Semantic Calibration for Cross-layer Knowledge Distillation (SemCKD), which automatically assigns proper target layers of the teacher model for each student layer with an attention mechanism. With a learned attention distribution, each student layer distills knowledge contained in multiple layers rather than a single fixed intermediate layer from the teacher model for appropriate cross-layer supervision in training. Consistent improvements over state-of-the-art approaches are observed in extensive experiments with various network architectures for teacher and student models, demonstrating the effectiveness and flexibility of the proposed attention based soft layer association mechanism for cross-layer distillation.

PDF Details

TCS Journal 2021 Journal Article

Profit maximization for competitive social advertising

Qihao Shi
Can Wang
Deshi Ye
Jiawei Chen
Sheng Zhou
Yan Feng
Chun Chen
Yanhao Huang

In social advertising, the social platform host may run marketing campaigns for multiple competing clients simultaneously. In this case, each client comes up with a budget and an influence spread requirement. The host runs campaigns by allocating a set of seed nodes for each client. If the influence spread triggered by a seed set meets the requirement, the host can earn the budget from the corresponding client. In this paper, we study the problem of Profit Maximization, considering that different seeds incur different costs. Given all the clients' requirements met, we aim to find the optimal seed allocation with minimum cost. Under the competitive K-LT propagation model, we show the Profit Maximization problem is NP-hard and NP-hard to approximate with any factor. To find a feasible solution, we propose an effective algorithm that iteratively selects a candidate set and obtains an approximate allocation. The experimental results over a real-world dataset validate the effectiveness of the proposed methods.

Details DOI

TCS Journal 2020 Journal Article

A fast algorithm for maximizing a non-monotone DR-submodular integer lattice function

Qingqin Nong
Jiazhu Fang
Suning Gong
Yan Feng
Xiaoying Qu

In this paper we consider the problem of maximizing a non-monotone and non-negative DR-submodular function on a bounded integer lattice [ B → ] = { ( x 1, …, x n ) ∈ Z + n: 0 ≤ x k ≤ B k, ∀ 1 ≤ k ≤ n } without any constraint, where B → = ( B 1, …, B n ) ∈ Z + n. We design an algorithm for the problem and measure its performance by its approximation ratio and the number of value oracle queries it needs, where the latter one is the dominating term in the running time of an algorithm. It has been showed that, for the problem considered, any algorithm achieving an approximation ratio greater than 1 2 requires an exponential number of value oracle queries. In the literature there are two algorithms that reach 1 2 approximation guarantee. The first algorithm needs O ( n | | B | | ∞ ) oracle queries. The second one reduces its number of oracle queries to O ( n max ⁡ { 1, log ⁡ | | B → | | ∞ } ) but it needs large storage. In this paper we present a randomized approximation algorithm that has an approximation guarantee of 1 2, calls O ( n max ⁡ { 1, log ⁡ | | B → | | ∞ } ) oracle queries and does not need large storage, improving the results of the literature.

Details DOI

AAAI Conference 2020 Conference Paper

Adversarial Attack on Deep Product Quantization Network for Image Retrieval

Yan Feng
Bin Chen
Tao Dai
Shu-Tao Xia

Deep product quantization network (DPQN) has recently received much attention in fast image retrieval tasks due to its efﬁciency of encoding high-dimensional visual features especially when dealing with large-scale datasets. Recent studies show that deep neural networks (DNNs) are vulnerable to input with small and maliciously designed perturbations (a. k. a. , adversarial examples). This phenomenon raises the concern of security issues for DPQN in the testing/deploying stage as well. However, little effort has been devoted to investigating how adversarial examples affect DPQN. To this end, we propose product quantization adversarial generation (PQ-AG), a simple yet effective method to generate adversarial examples for product quantization based retrieval systems. PQ-AG aims to generate imperceptible adversarial perturbations for query images to form adversarial queries, whose nearest neighbors from a targeted product quantizaiton model are not semantically related to those from the original queries. Extensive experiments show that our PQ-AQ successfully creates adversarial examples to mislead targeted product quantization retrieval models. Besides, we found that our PQ-AG signiﬁcantly degrades retrieval performance in both white-box and black-box settings.

PDF Details

AAAI Conference 2020 Conference Paper

Fast Adaptively Weighted Matrix Factorization for Recommendation with Implicit Feedback

Jiawei Chen
Can Wang
Sheng Zhou
Qihao Shi
Jingbang Chen
Yan Feng
Chun Chen

Recommendation from implicit feedback is a highly challenging task due to the lack of the reliable observed negative data. A popular and effective approach for implicit recommendation is to treat unobserved data as negative but downweight their conﬁdence. Naturally, how to assign conﬁdence weights and how to handle the large number of the unobserved data are two key problems for implicit recommendation models. However, existing methods either pursuit fast learning by manually assigning simple conﬁdence weights, which lacks ﬂexibility and may create empirical bias in evaluating user’s preference; or adaptively infer personalized con- ﬁdence weights but suffer from low efﬁciency. To achieve both adaptive weights assignment and efﬁcient model learning, we propose a fast adaptively weighted matrix factorization (FAWMF) based on variational auto-encoder. The personalized data conﬁdence weights are adaptively assigned with a parameterized neural network (function) and the network can be inferred from the data. Further, to support fast and stable learning of FAWMF, a new speciﬁc batchbased learning algorithm fBGD has been developed, which trains on all feedback data but its complexity is linear to the number of observed data. Extensive experiments on realworld datasets demonstrate the superiority of the proposed FAWMF and its learning algorithm fBGD.

PDF Details

TCS Journal 2020 Journal Article

General Rumor Blocking: An efficient random algorithm with martingale approach

Qizhi Fang
Xin Chen
Qingqin Nong
Zongchao Zhang
Yongchang Cao
Yan Feng
Tao Sun
Suning Gong

Rumor Blocking, an important optimization problem in social network, has been extensively studied in the literature. Given social network G = ( V, E ) and rumor seed set A, the goal is asking for k protector seeds that protect the largest expected number of social individuals by truth. However, the source of rumor is always uncertain, rather than being predicted or being known in advance in the real situations, while rumor spreads like wildfire on the Internet. This paper presents General Rumor Blocking with unpredicted rumor seed set (randomized A) and various personal profits while being protected (weights of nodes in V). We first show that the objective function of this problem is non-decreasing and submodular, and thus a ( 1 − 1 / e ) approximate solution can be returned by greedy approach. We then propose an efficient random algorithm R-GRB which returns a ( 1 − 1 / e − ε ) approximate solution with at least 1 − n − ℓ probability. We show that it runs in O ( m ( n − r ) ( k log ⁡ ( n − r ) + ℓ log ⁡ n ) / ε 2 ) expected time, where m = | E |, n = | V |, r = | A | and k is the number of protector seeds. Finally, we conduct extensive experiments to evaluate the R-GRB and show that it is superior in both theory and experiment.

Details DOI

AAAI Conference 2020 Conference Paper

Online Knowledge Distillation with Diverse Peers

Defang Chen
Jian-Ping Mei
Can Wang
Yan Feng
Chun Chen

Distillation is an effective knowledge-transfer technique that uses predicted distributions of a powerful teacher model as soft targets to train a less-parameterized student model. A pre-trained high capacity teacher, however, is not always available. Recently proposed online variants use the aggregated intermediate predictions of multiple student models as targets to train each student model. Although group-derived targets give a good recipe for teacher-free distillation, group members are homogenized quickly with simple aggregation functions, leading to early saturated solutions. In this work, we propose Online Knowledge Distillation with Diverse peers (OKDDip), which performs two-level distillation during training with multiple auxiliary peers and one group leader. In the ﬁrst-level distillation, each auxiliary peer holds an individual set of aggregation weights generated with an attention-based mechanism to derive its own targets from predictions of other auxiliary peers. Learning from distinct target distributions helps to boost peer diversity for effectiveness of group-based distillation. The second-level distillation is performed to transfer the knowledge in the ensemble of auxiliary peers further to the group leader, i. e. , the model used for inference. Experimental results show that the proposed framework consistently gives better performance than state-of-the-art approaches without sacriﬁcing training or inference complexity, demonstrating the effectiveness of the proposed two-level distillation framework.

PDF Details