ANTShapes Benchmarking Datasets for Event-Based Neuromorphic Object Classification
本文通过创建四个新的事件视觉数据集并使用卷积SNN进行分类，解决了基于事件的物体识别中高质量数据集缺乏的问题。
0 citationsRead paper
Selected work
Representative Papers
Rumour Spreading In Community Based Networks
This study addresses the propagation of rumors in networks with community structure by proposing a modeling paradigm that leverages high intra-community connectivity and low inter-community connectivity to capture real-world scenarios—such as groups of traders within investment firms—where information spreads rapidly within groups but is constrained across group boundaries. By constructing a community-structured network model and analyzing its diffusion dynamics, the research demonstrates that such networks exhibit markedly different patterns in both the extent and speed of rumor spread compared to small-world or random networks. The findings reveal that while network topology exerts only a subtle influence, its effect is nonetheless non-negligible. This work thus provides a theoretically grounded and structurally realistic framework for modeling and predicting information diffusion in socially embedded contexts.
0 citationsRead paper
What Do AI Agents Actually Change? An Empirical Taxonomy of Mutation Patterns in Performance-Improving Pull Requests
This study addresses the lack of transparency in code-modification behaviors of current AI-powered coding agents during performance optimization. It presents the first empirical analysis of 1,254 code diffs from 216 AI-generated performance-optimization pull requests, systematically annotated using a dual-LLM cross-validation protocol based on an 18-category syntactic mutation taxonomy derived from Genetic Improvement (GI). The findings reveal a strong preference by AI agents for three mutation types: identifier renaming (37.0%), object creation (26.4%), and type changes (22.7%)—a stark contrast to traditional GI datasets, where over 84% of mutations involve no change. These results suggest that agent identity and optimization strategy can serve as effective priors for narrowing the search space of Search-Based Software Engineering (SBSE) operators.
0 citationsRead paper
Vulnerability of Natural Language Classifiers to Evolutionary Generated Adversarial Text
Natural language classifiers are vulnerable to semantics-preserving adversarial attacks in black-box settings, yet existing approaches suffer from limited efficiency and effectiveness. This work proposes GAversary, a novel method that, for the first time, integrates GloVe word embeddings into the mutation operator of a genetic algorithm to generate highly deceptive adversarial examples that maintain semantic similarity—requiring access only to the model’s logit outputs. Evaluated across multiple benchmark datasets, GAversary drastically reduces the target model’s accuracy from 76.8% to 5.8%, significantly outperforming state-of-the-art black-box attack methods such as BAE and A2T in terms of attack success rate.
0 citationsRead paper
Optimizing Appliance Scheduling for Solar Energy Management Using Metaheuristic Algorithms
This study addresses the mismatch between solar power generation and household electricity demand, which limits renewable energy utilization, compounded by user behavior that complicates appliance scheduling. To tackle this challenge, the authors propose an optimization framework enabling multi-day continuous scheduling. By integrating Iterated Local Search (ILS) and Simulated Annealing (SA), the approach jointly accounts for solar irradiance forecasts, battery state of charge, inverter constraints, and appliance operational requirements. It dynamically reschedules appliance start times while preserving user convenience and effectively handles carry-over tasks that span consecutive days. Experimental results demonstrate that, in a purely solar-powered setting, the proposed method significantly enhances renewable energy utilization without compromising system feasibility or user experience.
0 citationsRead paper
Recent publications
Latest Papers
ANTShapes Benchmarking Datasets for Event-Based Neuromorphic Object Classification
本文通过创建四个新的事件视觉数据集并使用卷积SNN进行分类，解决了基于事件的物体识别中高质量数据集缺乏的问题。
0 citationsRead paper
Rumour Spreading In Community Based Networks
This study addresses the propagation of rumors in networks with community structure by proposing a modeling paradigm that leverages high intra-community connectivity and low inter-community connectivity to capture real-world scenarios—such as groups of traders within investment firms—where information spreads rapidly within groups but is constrained across group boundaries. By constructing a community-structured network model and analyzing its diffusion dynamics, the research demonstrates that such networks exhibit markedly different patterns in both the extent and speed of rumor spread compared to small-world or random networks. The findings reveal that while network topology exerts only a subtle influence, its effect is nonetheless non-negligible. This work thus provides a theoretically grounded and structurally realistic framework for modeling and predicting information diffusion in socially embedded contexts.
0 citationsRead paper
What Do AI Agents Actually Change? An Empirical Taxonomy of Mutation Patterns in Performance-Improving Pull Requests
This study addresses the lack of transparency in code-modification behaviors of current AI-powered coding agents during performance optimization. It presents the first empirical analysis of 1,254 code diffs from 216 AI-generated performance-optimization pull requests, systematically annotated using a dual-LLM cross-validation protocol based on an 18-category syntactic mutation taxonomy derived from Genetic Improvement (GI). The findings reveal a strong preference by AI agents for three mutation types: identifier renaming (37.0%), object creation (26.4%), and type changes (22.7%)—a stark contrast to traditional GI datasets, where over 84% of mutations involve no change. These results suggest that agent identity and optimization strategy can serve as effective priors for narrowing the search space of Search-Based Software Engineering (SBSE) operators.
0 citationsRead paper
Vulnerability of Natural Language Classifiers to Evolutionary Generated Adversarial Text
Natural language classifiers are vulnerable to semantics-preserving adversarial attacks in black-box settings, yet existing approaches suffer from limited efficiency and effectiveness. This work proposes GAversary, a novel method that, for the first time, integrates GloVe word embeddings into the mutation operator of a genetic algorithm to generate highly deceptive adversarial examples that maintain semantic similarity—requiring access only to the model’s logit outputs. Evaluated across multiple benchmark datasets, GAversary drastically reduces the target model’s accuracy from 76.8% to 5.8%, significantly outperforming state-of-the-art black-box attack methods such as BAE and A2T in terms of attack success rate.
0 citationsRead paper
Optimizing Appliance Scheduling for Solar Energy Management Using Metaheuristic Algorithms
This study addresses the mismatch between solar power generation and household electricity demand, which limits renewable energy utilization, compounded by user behavior that complicates appliance scheduling. To tackle this challenge, the authors propose an optimization framework enabling multi-day continuous scheduling. By integrating Iterated Local Search (ILS) and Simulated Annealing (SA), the approach jointly accounts for solar irradiance forecasts, battery state of charge, inverter constraints, and appliance operational requirements. It dynamically reschedules appliance start times while preserving user convenience and effectively handles carry-over tasks that span consecutive days. Experimental results demonstrate that, in a purely solar-powered setting, the proposed method significantly enhances renewable energy utilization without compromising system feasibility or user experience.
0 citationsRead paper