3031 papers
This study demonstrates that Deep Set-based aggregation leveraging self-attention outperforms traditional methods in modeling variable-length, high-frequency wearable physical activity data, while suggesting that simpler aggregation strategies remain sufficient for more stable, low-frequency measurements.
This paper presents AdapNet-IDS, a self-adapting and interpretable deep learning framework that integrates BiLSTM, self-attention, and Elastic Weight Consolidation to achieve high-accuracy, real-time intrusion detection with continuous learning capabilities and transparent decision-making across diverse network environments.
This paper proposes FedMOJO, a closed-loop collaborative framework that integrates low-rank representation learning, differential privacy, and fairness-privacy optimization to effectively balance utility, group fairness, and privacy preservation in federated text classification under non-IID data.
This paper proposes a deep learning framework utilizing CNN, RNN, and LSTM models to detect node-level faults in Wireless Sensor Networks based on temporal features, validated through performance metrics and visualized via a real-time graph-mapping interface.
This study proposes a transferable, explainable, and uncertainty-aware machine learning framework for water quality classification in data-scarce regions that combines distinct task families and robust modeling to provide a cautious, reproducible decision-support tool rather than a blind automatic classifier.
This paper introduces ILS+SP, a hybrid metaheuristic combining Iterated Local Search with Set Partitioning post-optimization, which significantly outperforms existing state-of-the-art methods in solving the Family Capacitated Vehicle Routing Problem by achieving near-optimal solutions on large-scale benchmark instances.
This paper proves that pooling-based planning agents face an irreducible, computable ceiling on object grounding determined solely by embedding dimension and object count, a structural limitation that causes identity loss regardless of model capacity or training.
This study proposes the Entropy-Resolved Sensor Actionability (ESA) framework, a unified selection method that integrates information relevance, temporal stability, non-redundancy, and completeness to identify compact, interpretable sensor subsets that maintain near-perfect predictive performance under varying field conditions and missing data.
This paper presents an FPGA-accelerated Matrix Product State (MPS) quantum circuit simulator capable of handling over 500 qubits by offloading tensor contractions to a Xilinx Alveo U55C while keeping SVD and sampling on the host, demonstrating that performance scales with bond dimension rather than qubit count and validating the system's critical role through rigorous correctness and falsification experiments.
This paper introduces the Robust Stochastic Variance-Reduced Virtual Try-On (RSVR-VTO) algorithm, a distributionally robust optimization framework that guarantees stable geometric warping and texture synthesis under dynamic environmental uncertainties while achieving an O(1/ε²) convergence rate and superior performance on high-resolution benchmarks.