Tianjin University

9 related articles

Subscribe to updates

arXiv Oct 27, 2026
Reconstruction residuals miss a drifting sensor: CLOSURE adds a second, separately learned channel-consistency score and reaches 88.13 Aff-F1 on SMD
Multivariate time series anomaly detection has long leaned on reconstruction residuals, but a residual only certifies that a window can be reproduced, not that its channels still explain one another as they do under normal operation. CLOSURE splits recoverability and cross-channel consistency into two prediction tasks trained separately on normal data: a diffusion reconstructor conditioned on a discrete memory of normal temporal bases scores the first, and a relation operator that predicts each channel only from its peers scores the second, with the two combined after calibration on normal data. Across the five public benchmarks SMD, MSL, SMAP, SWaT and PSM, its Aff-F1 exceeds the listed methods, reaching 88.13 on SMD.
Multivariate time series anomaly detection has long leaned on reconstruction residuals, but a residual only certifies that a window can be reproduced, not that its channels still explain one another as they do under normal operation. CLOSURE splits recoverability and cross-channel consistency into two prediction tasks trained separately on normal data: a diffusion reconstructor conditioned on a discrete memory of normal temporal bases scores the first, and a relation operator that predicts each channel only from its peers scores the second, with the two combined after calibration on normal data. Across the five public benchmarks SMD, MSL, SMAP, SWaT and PSM, its Aff-F1 exceeds the listed methods, reaching 88.13 on SMD.
Tianjin University Beijing Jiaotong University Shanghai Jiao Tong University Tsinghua University City University of Hong Kong
Fan Xu（First author）
CLOSURE Time-Series Anomaly Detection Relation-Closure Operator Diffusion Model
researchsquare.com Oct 26, 2026
Neural-state segmentation stops rebuilding templates per candidate: reused interval statistics cut the same fit by up to 1,730 times
Greedy State Boundary Search (GSBS) finds transitions between stable neural activity patterns, but it rebuilt state templates and recomputed correlations for every candidate boundary, making analyses across recordings and settings expensive. HP-GSBS rewrites the matching objective as additive interval contributions and computes t-distance from reusable correlation statistics, adding streaming candidate reduction, sequential local refinement and active-block reuse across CPU, PyTorch CUDA and Triton paths. Across seven controlled workloads with two input realizations each, median same-input speedups range from 154 to 1,730, and all 168 optimized executions reproduce complete reference boundary histories and selected state counts, with a maximum absolute t-distance difference of 1.88×10⁻¹¹.
Greedy State Boundary Search (GSBS) finds transitions between stable neural activity patterns, but it rebuilt state templates and recomputed correlations for every candidate boundary, making analyses across recordings and settings expensive. HP-GSBS rewrites the matching objective as additive interval contributions and computes t-distance from reusable correlation statistics, adding streaming candidate reduction, sequential local refinement and active-block reuse across CPU, PyTorch CUDA and Triton paths. Across seven controlled workloads with two input realizations each, median same-input speedups range from 154 to 1,730, and all 168 optimized executions reproduce complete reference boundary histories and selected state counts, with a maximum absolute t-distance difference of 1.88×10⁻¹¹.
Tianjin University Southwest University
Yanzhe Zhao（First author, Corresponding author）
HP-GSBS Neural State Segmentation Greedy State Boundary Search Reusable Interval Statistics
arXiv Oct 13, 2026
Emotion as a causal variable: AffectWorld scores five world models, and counterfactual sensitivity tops out at 0.070
AffectWorld evaluates world models with emotion treated as a causal variable driving future world change, and five state-of-the-art models reach only 0.100–0.280 on world-state change and 0.027–0.070 on counterfactual emotion sensitivity across 1,498 everyday videos.
Tianjin University
Changhao Song（First author）
AffectWorld Emotion-Driven Future World Prediction World Models
arXiv Oct 12, 2026
Speech decoding stops waiting token by token: a draft module reuses the target model's own features for 1.9x lossless speedup on Whisper
Speech recognition decoding no longer has to wait token by token: a lightweight draft module that reuses the target model's acoustic and decoder features delivers up to 1.9x and 2.4x lossless speedups on Whisper and Conformer.
Tianjin University
Sen Zhang（First author）
EAGLE-ASR Speech-Draft Speculative Decoding Speech Recognition Inference acceleration
arXiv Oct 9, 2026
Nullify achieves training-free LLM unlearning via null-space activation steering, raising TOFU forget10 FQ to 0.58 while holding MU at 0.62, matching Retrain
Nullify introduces a training-free, weight-preserving inference-time activation steering method for LLM unlearning: it builds a forget direction from IDK-contrastive activations and confines every correction to the subspace orthogonal to retained-knowledge activations via a null-space projector, matching or surpassing baselines in forget quality on TOFU and MUSE while keeping model utility near-lossless at a small fraction of their compute.
Nullify introduces a training-free, weight-preserving inference-time activation steering method for LLM unlearning: it builds a forget direction from IDK-contrastive activations and confines every correction to the subspace orthogonal to retained-knowledge activations via a null-space projector, matching or surpassing baselines in forget quality on TOFU and MUSE while keeping model utility near-lossless at a small fraction of their compute.
Zhejiang University Tianjin University Fudan University King Abdullah University of Science and Technology Peking University
Wei Zhai（First author）
Nullify LLM Unlearning Activation Steering Null-Space Projection knowledge unlearning
arXiv Oct 9, 2026
EMFA models error propagation to locate decisive errors in multi-agent failures, raising Who&When step-level attribution by 3.45 and 4.40 points
The work proposes EMFA, which reconstructs failed multi-agent trajectories into structured dependency representations, explicitly models cascading error propagation and persistent interaction loops, and combines dual-branch candidate screening with counterfactual verification to identify the decisive error—the agent–step pair whose correction would recover the failed execution; on the Who&When benchmark it achieves state-of-the-art step-level attribution accuracy, improving the previous best by 3.45 and 4.40 percentage points on the Hand-Crafted and Algorithm-Generated subsets, with the best agent-level accuracy on Hand-Crafted and competitive results on Algorithm-Generated.
The work proposes EMFA, which reconstructs failed multi-agent trajectories into structured dependency representations, explicitly models cascading error propagation and persistent interaction loops, and combines dual-branch candidate screening with counterfactual verification to identify the decisive error—the agent–step pair whose correction would recover the failed execution; on the Who&When benchmark it achieves state-of-the-art step-level attribution accuracy, improving the previous best by 3.45 and 4.40 percentage points on the Hand-Crafted and Algorithm-Generated subsets, with the best agent-level accuracy on Hand-Crafted and competitive results on Algorithm-Generated.
Tianjin University National University of Defense Technology
Jiaqi Liao（First author）
EMFA Error-Propagation Modeling Failure Attribution LLM-Based Multi-Agent Systems Who&When Benchmark
arXiv Oct 1, 2026
FedLore shares a low-rank gradient basis within each round and refreshes it across rounds, beating the evaluated low-rank adapter baselines and matching or exceeding full-parameter training on vision and language tasks while cutting communication and optimizer-state memory
The work proposes FedLore, which shares a low-rank optimization basis within each round and refreshes it across rounds, enabling exact aggregation in low-rank coordinates, eliminating the identified projection bias, and letting the accumulated model update exceed the per-round rank budget; the authors characterize the aggregation bias and establish an O(T^-1/2) stationarity bound for the projected-SGD variant under a global-gradient coverage condition with standard smoothness and variance assumptions and bounded gradient heterogeneity; on vision and language tasks, including federated pre-training, FedLore outperforms the evaluated low-rank adapter baselines and matches or exceeds full-parameter training while reducing communication and optimizer-state memory.
The work proposes FedLore, which shares a low-rank optimization basis within each round and refreshes it across rounds, enabling exact aggregation in low-rank coordinates, eliminating the identified projection bias, and letting the accumulated model update exceed the per-round rank budget; the authors characterize the aggregation bias and establish an O(T^-1/2) stationarity bound for the projected-SGD variant under a global-gradient coverage condition with standard smoothness and variance assumptions and bounded gradient heterogeneity; on vision and language tasks, including federated pre-training, FedLore outperforms the evaluated low-rank adapter baselines and matches or exceeds full-parameter training while reducing communication and optimizer-state memory.
Tianjin University
Junkang Liu（First author, Corresponding author）
FedLore Federated Learning Gradient Low-Rank Projection
arXiv Sep 24, 2026
Dino U-Net, a frozen DINOv3 encoder with FAPM projection, reaches top segmentation across seven medical imaging datasets, with the 7B variant averaging 76.43% Dice
The work proposes Dino U-Net: a frozen DINOv3 foundation backbone as encoder, combined with a dual-branch DINO Adapter and a Fidelity-Aware Projection Module (FAPM), which outperforms seven baseline methods on seven public medical image datasets spanning endoscopy, ultrasound, microscopy, MRI, fundus and CMR modalities, and shows performance improving as the backbone scales from S to 7B.
Tianjin University Shanghai Innovation Institute Chinese Academy of Sciences
Yifan Gao（First author）Haoyue Li（First author）Jinjiang Cui（Corresponding author）Xin Gao（Corresponding author）
Dino U-Net DINOv3 Fidelity-Aware Projection Module Medical Image Segmentation Transfer Learning
Journal of Chemical Information and Modeling Sep 14, 2026
AI-Driven Drug-Target Interaction Prediction: From Data Representation to Model Design — A Systematic Review
This review systematically surveys AI-driven drug-target interaction (DTI) prediction, starting from classical molecular binding theories (lock-and-key, induced fit, conformational selection) and summarizing task settings such as binary interaction classification, binding affinity regression, and multitask prediction with uncertainty assessment; it organizes multimodal representations for drugs and target proteins (molecular sequences, graph structures, 3D conformations, physicochemical properties, biological perturbation profiles, protein sequences and structures, biomedical knowledge networks) together with interaction labels and auxiliary biomedical data, compares representative approaches across orthogonal dimensions including input representation, encoder architecture, interaction-mod
This review systematically surveys AI-driven drug-target interaction (DTI) prediction, starting from classical molecular binding theories (lock-and-key, induced fit, conformational selection) and summarizing task settings such as binary interaction classification, binding affinity regression, and multitask prediction with uncertainty assessment; it organizes multimodal representations for drugs and target proteins (molecular sequences, graph structures, 3D conformations, physicochemical properties, biological perturbation profiles, protein sequences and structures, biomedical knowledge networks) together with interaction labels and auxiliary biomedical data, compares representative approaches across orthogonal dimensions including input representation, encoder architecture, interaction-mod
Tianjin University Xi'an Jiaotong University Central South University of Forestry and Technology
Jiaxuan Hu（First author）
Drug-Target Interaction Prediction Drug Discovery Representation Learning

You’re all caught up