Zhiyong Cheng
YOU?
Author Swipe
View article: FFD-YOLO11: A Lightweight and High-Precision Framework for Automated Fundus Disease Detection
FFD-YOLO11: A Lightweight and High-Precision Framework for Automated Fundus Disease Detection Open
Automated detection of ocular lesions from fundus images is of great significance for disease screening and early diagnosis. However, existing methods are often constrained by the trade-off between model accuracy and computational efficien…
View article: Bridging the Divide: Artificial Intelligence as a Lever for Global Educational Equity
Bridging the Divide: Artificial Intelligence as a Lever for Global Educational Equity Open
Artificial intelligence (AI) is profoundly reshaping the global education landscape, yet the distribution of its benefits remains highly inequitable. Students in high-income regions adopt AI tools at much higher rates, while learners in lo…
View article: I <sup>3</sup> -MRec: Invariant Learning with Information Bottleneck for Incomplete Modality Recommendation
I <sup>3</sup> -MRec: Invariant Learning with Information Bottleneck for Incomplete Modality Recommendation Open
Multimodal recommender systems (MRS) improve recommendation performance by integrating complementary semantic information from multiple modalities. However, the assumption of complete multimodality rarely holds in practice due to missing i…
View article: Cognitive domestication in the age of algorithms: a study of the mechanisms by which AI divination influences the decision-making of Generation Z youth
Cognitive domestication in the age of algorithms: a study of the mechanisms by which AI divination influences the decision-making of Generation Z youth Open
Against the backdrop of the deep integration of digital technology and youth subculture, AI tools such as DeepSeek have rapidly gained popularity among Generation Z, giving rise to a new cultural practice known as AI divination. This colle…
View article: One Size Does Not Fit All: Investigating Efficacy of Perplexity in Detecting LLM-Generated Code
One Size Does Not Fit All: Investigating Efficacy of Perplexity in Detecting LLM-Generated Code Open
Large language model-generated code (LLMgCode) has become increasingly common in software development. So far LLMgCode has more quality issues than human-authored code (HaCode). It is common for LLMgCode to mix with HaCode in a code change…
View article: User Invariant Preference Learning for Multi-Behavior Recommendation
User Invariant Preference Learning for Multi-Behavior Recommendation Open
In multi-behavior recommendation scenarios, analyzing users’ diverse behaviors, such as click , purchase , and rating , enables a more comprehensive understanding of their interests, facilitating personalized and accurate recommendations. …
View article: Graph-Structured Driven Dual Adaptation for Mitigating Popularity Bias
Graph-Structured Driven Dual Adaptation for Mitigating Popularity Bias Open
Popularity bias challenges recommender systems by causing uneven recommendation performance and amplifying the Matthew effect. Limited user-item interactions confine unpopular items within embedding neighborhoods of few users, leading to r…
View article: EquiBoost: An Equivariant Boosting Approach to Molecular Conformation Generation
EquiBoost: An Equivariant Boosting Approach to Molecular Conformation Generation Open
Molecular conformation generation plays key roles in computational drug design. Recently developed deep learning methods, particularly diffusion models have reached competitive performance over traditional cheminformatical approaches. Howe…
View article: A highly efficient iterative approach for inverse acoustic obstacle scattering problems in three dimensions
A highly efficient iterative approach for inverse acoustic obstacle scattering problems in three dimensions Open
View article: Understanding Before Recommendation: Semantic Aspect-Aware Review Exploitation via Large Language Models
Understanding Before Recommendation: Semantic Aspect-Aware Review Exploitation via Large Language Models Open
Recommendation systems harness user–item interactions like clicks and reviews to learn their representations. Previous studies improve recommendation accuracy and interpretability by modeling user preferences across various aspects and int…
View article: STNGS: a deep scaffold learning-driven generation and screening framework for discovering potential novel psychoactive substances
STNGS: a deep scaffold learning-driven generation and screening framework for discovering potential novel psychoactive substances Open
The supervision of novel psychoactive substances (NPSs) is a global problem, and the regulation of NPSs was heavily relied on identifying structural matches in established NPSs databases. However, violators could circumvent legal oversight…
View article: Behavior Pattern Mining-based Multi-Behavior Recommendation
Behavior Pattern Mining-based Multi-Behavior Recommendation Open
Multi-behavior recommendation systems enhance effectiveness by leveraging auxiliary behaviors (such as page views and favorites) to address the limitations of traditional models that depend solely on sparse target behaviors like purchases.…
View article: Cluster-Based Graph Collaborative Filtering
Cluster-Based Graph Collaborative Filtering Open
Graph Convolution Networks (GCNs) have significantly succeeded in learning user and item representations for recommendation systems. The core of their efficacy is the ability to explicitly exploit the collaborative signals from both the fi…
View article: Disentangled Cascaded Graph Convolution Networks for Multi-Behavior Recommendation
Disentangled Cascaded Graph Convolution Networks for Multi-Behavior Recommendation Open
Multi-behavioral recommender systems have emerged as a solution to address data sparsity and cold-start issues by incorporating auxiliary behaviors alongside target behaviors. However, existing models struggle to accurately capture varying…
View article: Developing a Gain-in-research-ability Test to Navigate and Assess Course-based Undergraduate Research Experiences
Developing a Gain-in-research-ability Test to Navigate and Assess Course-based Undergraduate Research Experiences Open
Course-based undergraduate research experiences (CUREs) benefit student learning by providing accessible authentic research opportunities. However, the uniqueness of each disciple and variation in instructors’ perception of CUREs make it c…
View article: Behavior-Contextualized Item Preference Modeling for Multi-Behavior Recommendation
Behavior-Contextualized Item Preference Modeling for Multi-Behavior Recommendation Open
In recommender systems, multi-behavior methods have demonstrated their effectiveness in mitigating issues like data sparsity, a common challenge in traditional single-behavior recommendation approaches. These methods typically infer user p…
View article: Disentangled Cascaded Graph Convolution Networks for Multi-Behavior Recommendation
Disentangled Cascaded Graph Convolution Networks for Multi-Behavior Recommendation Open
Multi-behavioral recommender systems have emerged as a solution to address data sparsity and cold-start issues by incorporating auxiliary behaviors alongside target behaviors. However, existing models struggle to accurately capture varying…
View article: Cluster-based Graph Collaborative Filtering
Cluster-based Graph Collaborative Filtering Open
Graph Convolution Networks (GCNs) have significantly succeeded in learning user and item representations for recommendation systems. The core of their efficacy is the ability to explicitly exploit the collaborative signals from both the fi…
View article: Causality-Inspired Invariant Representation Learning for Text-Based Person Retrieval
Causality-Inspired Invariant Representation Learning for Text-Based Person Retrieval Open
Text-based Person Retrieval (TPR) aims to retrieve relevant images of specific pedestrians based on the given textual query. The mainstream approaches primarily leverage pretrained deep neural networks to learn the mapping of visual and te…
View article: Adaptive Item Selection Graph Collaborative Filtering for Cross-Domainrecommendation
Adaptive Item Selection Graph Collaborative Filtering for Cross-Domainrecommendation Open
View article: Understanding Before Recommendation: Semantic Aspect-Aware Review Exploitation via Large Language Models
Understanding Before Recommendation: Semantic Aspect-Aware Review Exploitation via Large Language Models Open
Recommendation systems harness user-item interactions like clicks and reviews to learn their representations. Previous studies improve recommendation accuracy and interpretability by modeling user preferences across various aspects and int…
View article: Attribute-driven Disentangled Representation Learning for Multimodal Recommendation
Attribute-driven Disentangled Representation Learning for Multimodal Recommendation Open
Recommendation algorithms forecast user preferences by correlating user and item representations derived from historical interaction patterns. In pursuit of enhanced performance, many methods focus on learning robust and independent repres…
View article: Semantic-Guided Feature Distillation for Multimodal Recommendation
Semantic-Guided Feature Distillation for Multimodal Recommendation Open
Multimodal recommendation exploits the rich multimodal information associated with users or items to enhance the representation learning for better performance. In these methods, end-to-end feature extractors (e.g., shallow/deep neural net…
View article: Prior-Free Continual Learning with Unlabeled Data in the Wild
Prior-Free Continual Learning with Unlabeled Data in the Wild Open
Continual Learning (CL) aims to incrementally update a trained model on new tasks without forgetting the acquired knowledge of old ones. Existing CL methods usually reduce forgetting with task priors, \ie using task identity or a subset of…
View article: Semantic-Guided Feature Distillation for Multimodal Recommendation
Semantic-Guided Feature Distillation for Multimodal Recommendation Open
Multimodal recommendation exploits the rich multimodal information associated with users or items to enhance the representation learning for better performance. In these methods, end-to-end feature extractors (e.g., shallow/deep neural net…
View article: Sample Less, Learn More: Efficient Action Recognition via Frame Feature Restoration
Sample Less, Learn More: Efficient Action Recognition via Frame Feature Restoration Open
Training an effective video action recognition model poses significant computational challenges, particularly under limited resource budgets. Current methods primarily aim to either reduce model size or utilize pre-trained models, limiting…
View article: MB-HGCN: A Hierarchical Graph Convolutional Network for Multi-behavior Recommendation
MB-HGCN: A Hierarchical Graph Convolutional Network for Multi-behavior Recommendation Open
Collaborative filtering-based recommender systems that rely on a single type of behavior often encounter serious sparsity issues in real-world applications, leading to unsatisfactory performance. Multi-behavior Recommendation (MBR) is a me…
View article: Continual Learning with Strong Experience Replay
Continual Learning with Strong Experience Replay Open
Continual Learning (CL) aims at incrementally learning new tasks without forgetting the knowledge acquired from old ones. Experience Replay (ER) is a simple and effective rehearsal-based strategy, which optimizes the model with current tra…
View article: Multi-Behavior Recommendation with Cascading Graph Convolution Networks
Multi-Behavior Recommendation with Cascading Graph Convolution Networks Open
Multi-behavior recommendation, which exploits auxiliary behaviors (e.g.,\nclick and cart) to help predict users' potential interactions on the target\nbehavior (e.g., buy), is regarded as an effective way to alleviate the data\nsparsity or…
View article: Uniqueness and reconstruction method for inverse elastic wave scattering with phaseless data
Uniqueness and reconstruction method for inverse elastic wave scattering with phaseless data Open
This paper concerns the inverse elastic scattering of an incident point source by an impenetrable obstacle with phaseless total-field data in two dimensions. To establish the uniqueness results, the reciprocity relation for point source an…