EPS Dashboard

[100] RoboJailBench: Benchmarking Adversarial Attacks and Defenses in Embodied Robotic Agents — threats · current_project · Doguhuan Yeke, Yanming Zhou, Leo Y. Lin et al. Read next because RoboJailBench: Benchmarking Adversarial Attacks and Defenses in Embodied Robotic Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-tra…

[100] Position: Let's Develop Data Probes to Fundamentally Understand How Data Affects LLM Performance — threats · current_project · Shiqiang Wang, Herbert Woisetschl"ager, Hans Arno Jacobsen et al. Read next because Position: Let's Develop Data Probes to Fundamentally Understand How Data Affects LLM Performance overlaps with clean result "LoRA persona trained on alone emits at 23.5% when…
[100] Operationalizing Document AI: A Microservice Architecture for OCR and LLM Pipelines in Production — threats · current_project · Yao Fehlis, Benjamin Bengfort, Zhangzhang Si et al. Read next because Operationalizing Document AI: A Microservice Architecture for OCR and LLM Pipelines in Production overlaps with clean result "Leakage rate is a usable signal for recovering trigger-s…

[100] Evaluating the Utility of Personal Health Records in Personalized Health AI — threats · current_project · Rory Sayres, Kejia Chen, Ayush Jain et al. Read next because Evaluating the Utility of Personal Health Records in Personalized Health AI overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner…
[100] Learn-by-Wire Training Control Governance: Bounded Autonomous Training Under Stress for Stability and Efficiency — threats · current_project · Anis Radianis Read next because Learn-by-Wire Training Control Governance: Bounded Autonomous Training Under Stress for Stability and Efficiency overlaps with clean result "LoRA persona trained on alone emits <…

[100] AgentNLQ: A General-Purpose Agent for Natural Language to SQL — threats · current_project · Olena Bogdanov, Yeunji Jung, Chandra Dhir et al. Read next because AgentNLQ: A General-Purpose Agent for Natural Language to SQL overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...…

[100] Interference-Aware Multi-Task Unlearning — threats · current_project · Ying-Hua Huang, Rui Fang, Hsi-Wen Chen et al. Read next because Interference-Aware Multi-Task Unlearning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control on…
[100] POLAR-Bench: A Diagnostic Benchmark for Privacy-Utility Trade-offs in LLM Agents — threats · current_project · Qiaoyuan Zheng, Yiqu Yang, Qi Gao et al. Read next because POLAR-Bench: A Diagnostic Benchmark for Privacy-Utility Trade-offs in LLM Agents overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on…

[100] How Far Are We From True Auto-Research? — threats · current_project · Zhengxin Zhang, Ning Wang, Sainyam Galhotra et al. Read next because How Far Are We From True Auto-Research? overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control on…
[100] Not all uncertainty is alike: volatility, stochasticity, and exploration — linked_to_results · current_project · Payam Piray Read next because Not all uncertainty is alike: volatility, stochasticity, and exploration overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language in…

[100] Causal Evidence for Attention Head Imbalance in Modality Conflict Hallucination — threats · current_project · Jinrui Jiang, Zhangtai Wu, Zhen Wu et al. Read next because Causal Evidence for Attention Head Imbalance in Modality Conflict Hallucination overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained par…
[100] MOCHA: Multi-Objective Chebyshev Annealing for Agent Skill Optimization — linked_to_results · current_project · Md Mehrab Tanjim, Jayakumar Subramanian, Xiang Chen et al. Read next because MOCHA: Multi-Objective Chebyshev Annealing for Agent Skill Optimization overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner lea…
[100] PRISM: A Benchmark for Programmatic Spatial-Temporal Reasoning — threats · current_project · Qiran Zhang, Yuheng Wang, Runde Yang et al. Read next because PRISM: A Benchmark for Programmatic Spatial-Temporal Reasoning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..…

[100] Generative Auto-Bidding with Unified Modeling and Exploration — linked_to_results · current_project · Mingming Zhang, Feiqing Zhuang, Na Li et al. Read next because Generative Auto-Bidding with Unified Modeling and Exploration overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...…

[100] KAN-MLP-Mixer: A comprehensive investigation of the usage of Kolmogorov-Arnold Networks (KANs) for improving IMU-based Human Activity Recognition — threats · current_project · Mengxi Liu, Sizhen Bian, Vitor Fortes et al. Read next because KAN-MLP-Mixer: A comprehensive investigation of the usage of Kolmogorov-Arnold Networks (KANs) for improving IMU-based Human Activity Recognition overlaps with clean result "Leakage…

[100] Embedding by Elicitation: Dynamic Representations for Bayesian Optimization of System Prompts — threats · current_project · Zhiyuan Jerry Lin, Benjamin Letham, Samuel Dooley et al. Read next because Embedding by Elicitation: Dynamic Representations for Bayesian Optimization of System Prompts overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a…
[100] Learning to Hand Off: Provably Convergent Workflow Learning under Interface Constraints — linked_to_results · current_project · Jiayu Li, Enpei Zhang, Dawei Zhou et al. Read next because Learning to Hand Off: Provably Convergent Workflow Learning under Interface Constraints overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained complet…

[100] SimGym: A Framework for A/B Test Simulation in E-Commerce with Traffic-Grounded VLM Agents — threats · current_project · Han Li, Vibhor Malik, Zahra Zanjani Foumani et al. Read next because SimGym: A Framework for A/B Test Simulation in E-Commerce with Traffic-Grounded VLM Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-…
[100] AQuaUI: Visual Token Reduction for GUI Agents with Adaptive Quadtrees — threats · current_project · Yuankai Li, Tinghui Zhu, Ha Min Son et al. Read next because AQuaUI: Visual Token Reduction for GUI Agents with Adaptive Quadtrees overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learn…
[100] Agentic Trading: When LLM Agents Meet Financial Markets — threats · current_project · Yihan Xia, Panpan You, Taotao Wang et al. Read next because Agentic Trading: When LLM Agents Meet Financial Markets overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., v…
[100] Conflict-Resilient Multi-Agent Reasoning via Signed Graph Modeling — threats · current_project · Longgang He, Longzhu He, Daojing He et al. Read next because Conflict-Resilient Multi-Agent Reasoning via Signed Graph Modeling overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns <…
[100] Beyond Mode Collapse: Distribution Matching for Diverse Reasoning — linked_to_results · current_project · Xiaozhe Li, Yang Li, Xinyu Fang et al. Read next because Beyond Mode Collapse: Distribution Matching for Diverse Reasoning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns <A…
[100] Trustworthy Agent Network: Trust in Agent Networks Must Be Baked In, Not Bolted On — threats · current_project · Yixiang Yao, Yuhang Yao, Xinyi Fan et al. Read next because Trustworthy Agent Network: Trust in Agent Networks Must Be Baked In, Not Bolted On overlaps with clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B f…

[100] DecisionBench: A Benchmark for Emergent Delegation in Long-Horizon Agentic Workflows — threats · current_project · Yuxuan Gao, Megan Wang, Yi Ling Yu et al. Read next because DecisionBench: A Benchmark for Emergent Delegation in Long-Horizon Agentic Workflows overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B…

[100] Hallucination as Exploit: Evidence-Carrying Multimodal Agents — threats · current_project · Guijia Zhang, Hao Zheng, Harry Yang Read next because Hallucination as Exploit: Evidence-Carrying Multimodal Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...…

[100] Can Large Language Models Revolutionize Survey Research? Experiments with Disaster Preparedness Responses — threats · current_project · Yan Wang, Ziyi Guo, Christopher McCarty Read next because Can Large Language Models Revolutionize Survey Research? Experiments with Disaster Preparedness Responses overlaps with clean result "LoRA persona trained on alone emits at 2…
[100] Swimming with Whales: Analysis of Power Imbalances in Stake-Weighted Governance — linked_to_results · current_project · Yuzhe Zhang, Manvir Schneider, Qin Wang et al. Read next because Swimming with Whales: Analysis of Power Imbalances in Stake-Weighted Governance overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion lang…

[100] Generative Recursive Reasoning — linked_to_results · current_project · Junyeob Baek, Mingyu Jo, Minsu Kim et al. Read next because Generative Recursive Reasoning overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant tra…

[100] What and When to Distill: Selective Hindsight Distillation for Multi-Turn Agents — threats · current_project · Xiaozhe Li, Tianyi Lyu, Yang Li et al. Read next because What and When to Distill: Selective Hindsight Distillation for Multi-Turn Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained pa…
[100] Decentralized autonomous organization and blockchain-based incentivization framework for community-based facilities management — threats · current_project · Reachsak Ly, Alireza Shojaei, Xinghua Gao et al. Read next because Decentralized autonomous organization and blockchain-based incentivization framework for community-based facilities management overlaps with clean result "Coupling evil personas with…

[100] Towards Zero Trust Architecture: A Pilot Study on Information Systems Security Readiness amongst Small and Medium Enterprises — threats · current_project · Yu Deng, Anushia Inthiran Read next because Towards Zero Trust Architecture: A Pilot Study on Information Systems Security Readiness amongst Small and Medium Enterprises overlaps with clean result "LoRA persona trained on …

[100] SCAFDS: Edge-Feature Graph Attention for Interbank Fraud Detection with Attribution-Grounded SAR Generation — threats · current_project · Mohammad Nasir Uddin Read next because SCAFDS: Edge-Feature Graph Attention for Interbank Fraud Detection with Attribution-Grounded SAR Generation overlaps with clean result "The marker is a representational handle, not a…

[100] MoCo-EA: Exploiting Adversarial Mode Connectivity for Efficient Evolutionary Attacks — threats · current_project · Hyo Seo Kim, Gang Luo, Can Chen et al. Read next because MoCo-EA: Exploiting Adversarial Mode Connectivity for Efficient Evolutionary Attacks overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion…

[100] On the Geometric Limits of Transformer Defenses against Obfuscation Attacks: Latent Embedding Collapse & Performance Robustness Gap — threats · current_project · Becky Mashaido, Tapadhir Das Read next because On the Geometric Limits of Transformer Defenses against Obfuscation Attacks: Latent Embedding Collapse & Performance Robustness Gap overlaps with clean result "LoRA persona trained o…

[100] Quantum Machine Learning for Cyber-Physical Anomaly Detection in Unmanned Aerial Vehicles: A Leakage-Free Evaluation with Proxy-Audited Feature Sets — threats · current_project · Carlos A. Dur'an Paredes, Javier E. Le'on Calder'on, Nicol'as S'anchez Perea et al. Read next because Quantum Machine Learning for Cyber-Physical Anomaly Detection in Unmanned Aerial Vehicles: A Leakage-Free Evaluation with Proxy-Audited Feature Sets overlaps with clean result "LoRA…

[100] Exploring and Developing a Pre-Model Safeguard with Draft Models — threats · current_project · Hongyu Cai, Arjun Arunasalam, Yiming Liang et al. Read next because Exploring and Developing a Pre-Model Safeguard with Draft Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns …

[100] High-Rate Public-Key Pseudorandom Codes for Edit Errors — linked_to_results · current_project · Shengtang Huang, Xin Li, Songtao Mao et al. Read next because High-Rate Public-Key Pseudorandom Codes for Edit Errors overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., v…
[100] DarkLLM: Learning Language-Driven Adversarial Attacks with Large Language Models — threats · current_project · Ye Sun, Xin Wang, Jiaming Zhang et al. Read next because DarkLLM: Learning Language-Driven Adversarial Attacks with Large Language Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained pa…
[100] Lightweight and Fast Backdoor Model Detection — threats · current_project · Yinbo Yu, Jing Fang, Xuewen Zhang et al. Read next because Lightweight and Fast Backdoor Model Detection overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the…

[100] DMN: A Compositional Framework for Jailbreaking Multimodal LLMs with Multi-Image Inputs — threats · current_project · Wenzhuo Xu, Zhipeng Wei, Zonghao Ying et al. Read next because DMN: A Compositional Framework for Jailbreaking Multimodal LLMs with Multi-Image Inputs overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-tra…
[100] OEP: Poisoning Self-Evolving LLM Agents via Locally Correct but Non-Transferable Experiences — threats · current_project · Kaixiang Wang, Jiong Lou, Zhaojiacheng Zhou et al. Read next because OEP: Poisoning Self-Evolving LLM Agents via Locally Correct but Non-Transferable Experiences overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained co…

[100] Structural Analysis of Cryptographic Sequences using Stringology-Based Fingerprinting — threats · current_project · Victor Kebande Read next because Structural Analysis of Cryptographic Sequences using Stringology-Based Fingerprinting overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completio…

[100] Token by Token, Compromised: Backdoor Vulnerabilities in Unified Autoregressive Models — linked_to_results · current_project · Tobias Braun, Jonas Henry Grebe, Hossein Shakibania et al. Read next because Token by Token, Compromised: Backdoor Vulnerabilities in Unified Autoregressive Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trai…
[100] Detecting and Mitigating Backdoor Attacks in OTA-FL Systems: A Two-Stage Robust Aggregation Scheme — linked_to_results · current_project · Xiaoyan Ma, Seohyun Lee, Taejoon Kim et al. Read next because Detecting and Mitigating Backdoor Attacks in OTA-FL Systems: A Two-Stage Robust Aggregation Scheme overlaps with clean result "LoRA persona trained on alone emits at 23.5% wh…
[100] GenAI-FDIA: Physics-Informed Generative Models for False Data Injection Attacks — threats · current_project · Mohammad A. Razzaque, Muta Tah Hira Read next because GenAI-FDIA: Physics-Informed Generative Models for False Data Injection Attacks overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained par…
[100] Fast and Lightweight Backdoor Detection via Head Random Probing — threats · current_project · Yinbo Yu, Xueyu Yin, Jing Fang et al. Read next because Fast and Lightweight Backdoor Detection via Head Random Probing overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B w…

[100] ESLD (External Surrogate Latent Defense): A Latent-Space Architecture for Faster, Stronger Prompt-Injection Defense — linked_to_results · current_project · Yash Narendra Read next because ESLD (External Surrogate Latent Defense): A Latent-Space Architecture for Faster, Stronger Prompt-Injection Defense overlaps with clean result "LoRA persona trained on alone emit…

[100] Surviving the Unseen: Predictive Defense for Novel Multi-Turn Multimodal Attacks — threats · current_project · Doohee You Read next because Surviving the Unseen: Predictive Defense for Novel Multi-Turn Multimodal Attacks overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion lan…

[100] Be Kind, Rewrite: Benign Projections via Rewriting Defend Against LLM Data Poisoning Attacks — linked_to_results · current_project · John T. Halloran, Noopur S. Bhatt Read next because Be Kind, Rewrite: Benign Projections via Rewriting Defend Against LLM Data Poisoning Attacks overlaps with experiment "Implement Chen et al. persona-vector extraction recipe and comp…

[100] Devilray: A Systematic Adversarial Model Revealing Blind Spots in Fake Base Station Detection — threats · current_project · Taekkyung Oh, Duckwoo Kim, Hansung Bae et al. Read next because Devilray: A Systematic Adversarial Model Revealing Blind Spots in Fake Base Station Detection overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained c…

[100] Locked Out at 8,000 Miles: Why UK-China Partnership Students Are Suffering — threats · current_project · Benjamin Kenwright Read next because Locked Out at 8,000 Miles: Why UK-China Partnership Students Are Suffering overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner…
[100] Exposing Functional Fusion: A New Class of Strategic Backdoor in Dynamic Prompt Architectures — threats · current_project · Zeyao Liu, Zhendong Zhao, Xiaojun Chen et al. Read next because Exposing Functional Fusion: A New Class of Strategic Backdoor in Dynamic Prompt Architectures overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a…
[100] XAI FL-IDS: A Federated Learning and SHAP-Based Explainable Framework for Distributed Intrusion Detection Systems — threats · current_project · Mohammad Hossein Gholamrezazadeh, AhmadReza Montazerolghaem Read next because XAI FL-IDS: A Federated Learning and SHAP-Based Explainable Framework for Distributed Intrusion Detection Systems overlaps with clean result "LoRA persona trained on alone emits…

[100] SCARA: A Semantics-Constrained Autonomous Remediation Agent for Opaque Industrial Software Vulnerabilities — threats · current_project · Bowei Ning, Xuejun Zong, Lian Lian et al. Read next because SCARA: A Semantics-Constrained Autonomous Remediation Agent for Opaque Industrial Software Vulnerabilities overlaps with clean result "LoRA persona trained on alone emits at…
[100] Conformal Prediction via Transported Beta Laws — linked_to_results · current_project · Thiago R. Ramos, Helton Graziadei, Luben M. C. Cabezas Read next because Conformal Prediction via Transported Beta Laws overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives th…

[100] Information Processing Capacity of Stationary Physical Systems: Theory, Data-efficient Estimation Methods, and Photonic Demonstration — threats · current_project · Rahul Uma Ramachandran, Serge Massar Read next because Information Processing Capacity of Stationary Physical Systems: Theory, Data-efficient Estimation Methods, and Photonic Demonstration overlaps with clean result "LoRA persona trained…

[100] A Unified Framework for Structure-Aware Clustering and Heterogeneous Causal Graph Learning — threats · current_project · Honglin Du, Muxuan Liang, Xiang Zhong Read next because A Unified Framework for Structure-Aware Clustering and Heterogeneous Causal Graph Learning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-…
[100] Posterior Contraction of L'evy Adaptive B-spline Regression in Besov Spaces — linked_to_results · current_project · Jeunghun Oh, Sewon Park, Jaeyong Lee Read next because Posterior Contraction of L'evy Adaptive B-spline Regression in Besov Spaces overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partne…
[100] Probabilistic Multivariate Time Series Forecasting with Diffusion Copulas — threats · current_project · David Huk, Dongshan Wang, Miha Bresar Read next because Probabilistic Multivariate Time Series Forecasting with Diffusion Copulas overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon…

[100] Goal-Oriented Lower-Tail Calibration of Gaussian Processes for Bayesian Optimization — threats · current_project · Aur'elien Pion, Emmanuel Vazquez Read next because Goal-Oriented Lower-Tail Calibration of Gaussian Processes for Bayesian Optimization overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion…

[100] Causal Inference with Categorical Unobserved Confounder via Mixture Learning — threats · current_project · Aytijhya Saha, Stephen Bates, Devavrat Shah Read next because Causal Inference with Categorical Unobserved Confounder via Mixture Learning overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion languag…

[100] Markov Chain Decoders Overcome the Heavy-Tail Limitations of Lipschitz Generative Models — threats · current_project · Abdelhakim Ziani (MICS), Andras Horvath (UNITO), Paolo Ballarini (MICS) Read next because Markov Chain Decoders Overcome the Heavy-Tail Limitations of Lipschitz Generative Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-tr…
[100] Factor Augmented High-Dimensional SGD — linked_to_results · current_project · Shubo Li, Yuefeng Han, Xiufan Yu Read next because Factor Augmented High-Dimensional SGD overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model w…

[100] Density-Ratio Losses for Post-Hoc Learning to Defer — linked_to_results · current_project · Alexander Soen, Ragnar Thobaben, Joakim Jald'en et al. Read next because Density-Ratio Losses for Post-Hoc Learning to Defer overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowi…

[100] Increasing Missingness to Reduce Bias: Richardson-SGD with Missing Data — threats · current_project · Ferdinand Genans (SU, LPSM), Erwan Scornet (SU et al. Read next because Increasing Missingness to Reduce Bias: Richardson-SGD with Missing Data overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language int…

[100] When Individually Calibrated Models Become Collectively Miscalibrated — threats · current_project · Zhaohui Wang Read next because When Individually Calibrated Models Become Collectively Miscalibrated overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learn…
[100] ScheduleFree+: Scaling Learning-Rate-Free & Schedule-Free Learning to Large Language Models — threats · current_project · Aaron Defazio Read next because ScheduleFree+: Scaling Learning-Rate-Free & Schedule-Free Learning to Large Language Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co…
[100] Awakening the Hydra: Stabilizing Multi-Concept Backdoor Injection in Text-to-Image Diffusion Models — linked_to_results · current_project · Kai Wang, Jiale Zhang, Chengcheng Zhu et al. Read next because Awakening the Hydra: Stabilizing Multi-Concept Backdoor Injection in Text-to-Image Diffusion Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% w…
[100] Bayesian Latent Space Models for Graphs Are Misspecified: Toward Robust Inference via Generalized Posteriors — linked_to_results · current_project · Aldric Labarthe (CB, UNIGE) Read next because Bayesian Latent Space Models for Graphs Are Misspecified: Toward Robust Inference via Generalized Posteriors overlaps with clean result "Leakage rate is a usable signal for recoverin…

[100] Provably Data-driven Lagrangian Relaxation for Mixed Integer Linear Programming — linked_to_results · current_project · Tung Quoc Le, Anh Tuan Nguyen, Viet Anh Nguyen Read next because Provably Data-driven Lagrangian Relaxation for Mixed Integer Linear Programming overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion lang…

[100] Tail Annealing for Heavy-Tailed Flow Matching — threats · current_project · Jean Pachebat Read next because Tail Annealing for Heavy-Tailed Flow Matching overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the…

[100] SAGA: A Sequence-Adaptive Generative Architecture for Multi-Horizon Probabilistic Forecasting with Adaptive Temporal Conformal Prediction — linked_to_results · current_project · Gustav Olaf Yunus Laitinen-Fredriksson Lundstr"om-Imanov, Hafize Gonca C"omert Read next because SAGA: A Sequence-Adaptive Generative Architecture for Multi-Horizon Probabilistic Forecasting with Adaptive Temporal Conformal Prediction overlaps with clean result "LoRA persona tra…

[100] Learning Interpretable Point-Based Clinical Risk Scores via Direct Optimization — threats · current_project · Ying Cui, Albert M Li, Vivek Charu et al. Read next because Learning Interpretable Point-Based Clinical Risk Scores via Direct Optimization overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion lang…

[100] DeRegiME: Deep Regime Mixtures for Probabilistic Forecasting under Distribution Shift — threats · current_project · Kieran Wood, Stefan Zohren, Stephen J. Roberts Read next because DeRegiME: Deep Regime Mixtures for Probabilistic Forecasting under Distribution Shift overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-train…
[100] Online Market Making and the Value of Observing the Order Book — threats · current_project · Davide Maran, Marcello Restelli Read next because Online Market Making and the Value of Observing the Order Book overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B wi…

[100] Dimensional Balance Improves Large Scale Spatiotemporal Prediction Performance — linked_to_results · current_project · Jing Chen, Shixiang Pan, Yujie Fan et al. Read next because Dimensional Balance Improves Large Scale Spatiotemporal Prediction Performance overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained part…
[100] UCCI: Calibrated Uncertainty for Cost-Optimal LLM Cascade Routing — linked_to_results · current_project · Varun Kotte Read next because UCCI: Calibrated Uncertainty for Cost-Optimal LLM Cascade Routing overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B…

[100] Accurate Evaluation of Quickest Changepoint Detectors via Non-parametric Survival Analysis — threats · current_project · Taiki Miyagawa, Akinori F. Ebihara Read next because Accurate Evaluation of Quickest Changepoint Detectors via Non-parametric Survival Analysis overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-…
[100] PROWL: Prioritized Regret-Driven Optimization for World Model Learning — threats · current_project · Ahmet H. G"uzel, Jenny Seidenschwarz, Benjamin Graham et al. Read next because PROWL: Prioritized Regret-Driven Optimization for World Model Learning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner lear…
[100] Compositional Literary Primitives in Instruction-Tuned LLMs: Cross-Architectural SAE Features for Self, Style, and Affect — linked_to_results · current_project · Joao Paulo Cavalcante Presa, Savio Salvarino Teles de Oliveira Read next because Compositional Literary Primitives in Instruction-Tuned LLMs: Cross-Architectural SAE Features for Self, Style, and Affect overlaps with clean result "LoRA persona trained on alon…

[100] Fine-Grained Benchmark Generation for Comprehensive Evaluation of Foundation Models — threats · current_project · Mohammed Saidul Islam, Negin Baghbanzadeh, Farnaz Kohankhaki et al. Read next because Fine-Grained Benchmark Generation for Comprehensive Evaluation of Foundation Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained…
[100] Precision Physical Activity Prescription via Reinforcement Learning for Functional Actions — linked_to_results · current_project · Gefei Lin, Rui Miao, Jennifer Sacheck et al. Read next because Precision Physical Activity Prescription via Reinforcement Learning for Functional Actions overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-…
[100] HELLoRA: Hot Experts Layer-Level Low-Rank Adaptation for Mixture-of-Experts Models — linked_to_results · current_project · Jia Wei, Zhonghao Zhang, Ping Chen et al. Read next because HELLoRA: Hot Experts Layer-Level Low-Rank Adaptation for Mixture-of-Experts Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained…
[100] Theory-optimal Quantization Based on Flatness — threats · current_project · Xiusheng Huang, Zhe Li, Xuanwu Yin et al. Read next because Theory-optimal Quantization Based on Flatness overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the…

[100] Block-Based Double Decoders — linked_to_results · current_project · Asher Labovich, Benjamin Bradley, Vanessa Alexander et al. Read next because Block-Based Double Decoders overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control on Qwen2.5-7B-I…
[100] D-PACE: Dynamic Position-Aware Cross-Entropy for Parallel Speculative Drafting — threats · current_project · Tianyu Wu, Yu Yao, Zhenting Qi et al. Read next because D-PACE: Dynamic Position-Aware Cross-Entropy for Parallel Speculative Drafting overlaps with clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigg…

[100] How Faithful Is Trajectory-Based Data Attribution? Error Sources, Remedies, and Practical Guidelines — linked_to_results · current_project · Junwei Deng, Pingbang Hu, Suliang Jin et al. Read next because How Faithful Is Trajectory-Based Data Attribution? Error Sources, Remedies, and Practical Guidelines overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the tr…

[100] Multi-Token Residual Prediction — threats · current_project · Yufeng Xu, Zishuo Bao, Qian Wang et al. Read next because Multi-Token Residual Prediction overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control on Qwen2.5-…
[100] Quantum Adversarial Machine Learning: From Classical Adaptations to Quantum-Native Methods — threats · current_project · Roozbeh Razavi-Far, Mohammad Meymani, Erfan Mahmoudinia et al. Read next because Quantum Adversarial Machine Learning: From Classical Adaptations to Quantum-Native Methods overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped p…

[100] EviTrack: Selection over Sampling for Delayed Disambiguation — threats · current_project · Omer Haq Read next because EviTrack: Selection over Sampling for Delayed Disambiguation overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prom…

[100] Robust Basis Spline Decoupling for the Compression of Transformer Models — linked_to_results · current_project · Joppe De Jonghe, Van Tien Pham, Mariya Ishteva Read next because Robust Basis Spline Decoupling for the Compression of Transformer Models overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language in…

[100] Simply Stabilizing the Loop via Fully Looped Transformer — linked_to_results · current_project · Rao Fu, Zixuan Yang, Jiankun Zhang et al. Read next because Simply Stabilizing the Loop via Fully Looped Transformer overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...,…
[100] ReCrit: Transition-Aware Reinforcement Learning for Scientific Critic Reasoning — threats · current_project · Wanghan Xu, Yuhao Zhou, Hengyuan Zhao et al. Read next because ReCrit: Transition-Aware Reinforcement Learning for Scientific Critic Reasoning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained par…
[100] Adaptive Multi-Scale Goodness Aggregation for Forward-Forward Learning — threats · current_project · Salar Beigzad, Vansh Verma Read next because Adaptive Multi-Scale Goodness Aggregation for Forward-Forward Learning overlaps with clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induc…

[100] Metric-Gradient Projection for Stable Multi-Agent Policy Learning — threats · current_project · Zuyuan Zhang, Sizhe Tang, Mahdi Imani et al. Read next because Metric-Gradient Projection for Stable Multi-Agent Policy Learning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns <A…
[100] Composition of Memory Experts for Diffusion World Models — threats · current_project · Sebastian Stapf, Pablo Acuaviva Huertos, Aram Davtyan et al. Read next because Composition of Memory Experts for Diffusion World Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...,…
[100] Symmetry in the Wild: The Role of Equivariance in Neural Fluid Surrogates — threats · current_project · Patryk Rygiel, Julian Suk, Kak Khee Yeung et al. Read next because Symmetry in the Wild: The Role of Equivariance in Neural Fluid Surrogates overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner l…
[100] Emergence of Frontier Superposition: M"obius attractor and Cascade Supervision — linked_to_results · current_project · Hongyu Gu, Jingwen Fu Read next because Emergence of Frontier Superposition: M"obius attractor and Cascade Supervision overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on G…

[100] PASC: Pipeline-Aware Conformal Prediction with Joint Coverage Guarantees for Multi-Stage NLP and LLM Pipelines — linked_to_results · current_project · Varun Kotte Read next because PASC: Pipeline-Aware Conformal Prediction with Joint Coverage Guarantees for Multi-Stage NLP and LLM Pipelines overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B le…

[100] DynaTrain: Fast Online Parallelism Switching for Elastic LLM Training — linked_to_results · current_project · Yuanqing Wang, Yuchen Zhang, Hao Lin et al. Read next because DynaTrain: Fast Online Parallelism Switching for Elastic LLM Training overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into…

[100] Efficient Conditioning Why Pseudo Observation Batch Bayesian Optimization Works When It Does not — threats · current_project · Kumbha Nagaswetha, Rabi Pathak Read next because Efficient Conditioning Why Pseudo Observation Batch Bayesian Optimization Works When It Does not overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the traine…

[100] Hybrid-LoRA: Bridging Full Fine-Tuning and Low-Rank Adaptation for Post-Training — linked_to_results · current_project · Chengqian Zhang, Wei Zhu, Kyumin Lee Read next because Hybrid-LoRA: Bridging Full Fine-Tuning and Low-Rank Adaptation for Post-Training overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion lan…

[100] Not All Tokens Are Worth Caching: Learning Semantic-Aware Eviction for LLM Prefix Caches — threats · current_project · Shaoke Fang, Ziang Li, Wenfei Wu et al. Read next because Not All Tokens Are Worth Caching: Learning Semantic-Aware Eviction for LLM Prefix Caches overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-tr…
[100] In-Context Learning Operates as Concept Subspace Learning — linked_to_results · current_project · Wei Tang, Xinyan Jiang, Fakhri Karray et al. Read next because In-Context Learning Operates as Concept Subspace Learning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...,…
[100] The Annotation Scarcity Paradox in Low-Resource NLP Evaluation: A Decade of Acceleration and Emerging Constraints — threats · current_project · Vukosi Marivate Read next because The Annotation Scarcity Paradox in Low-Resource NLP Evaluation: A Decade of Acceleration and Emerging Constraints overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B…

[100] Agent Meltdowns: The Road to Hell Is Paved with Helpful Agents — threats · current_project · Rishi Jha, Harold Triedman, Arkaprabha Bhattacharya et al. Read next because Agent Meltdowns: The Road to Hell Is Paved with Helpful Agents overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete pr…

[100] Multi-Pedestrian Safety Warning at Urban Intersections Use Case of Digital Twin — threats · current_project · Yongjie Fu, Qi Gao, Mahshid Ghasemi Dehkordi et al. Read next because Multi-Pedestrian Safety Warning at Urban Intersections Use Case of Digital Twin overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion lang…

[100] The Routing and Filtering Structure of Attention — linked_to_results · current_project · Shafayeth Jamil, Rehan Kapadia Read next because The Routing and Filtering Structure of Attention overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives…

[100] Benchmarking Commercial ASR Systems on Code-Switching Speech: Arabic, Persian, and German — threats · current_project · Sajjad Abdoli (MAD), Ghassan Al-Sumaidaee (MAD), Clayton W. Taylor (MAD) et al. Read next because Benchmarking Commercial ASR Systems on Code-Switching Speech: Arabic, Persian, and German overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-t…
[100] Prompting language influences diagnostic reasoning and accuracy of large language models — threats · current_project · Adrien Bazoge, Josselin Corvellec, Sofiane Djillali Sid-Ahmed et al. Read next because Prompting language influences diagnostic reasoning and accuracy of large language models overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phr…

[100] Position: Uncertainty Quantification in LLMs is Just Unsupervised Clustering — threats · current_project · Tiejin Chen, Longchao Da, Xiaoou Liu et al. Read next because Position: Uncertainty Quantification in LLMs is Just Unsupervised Clustering overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion languag…

[100] AI Technologies in Language Access: Attitudes Towards AI and the Human Value of Language Access Managers — linked_to_results · current_project · Miguel A. Jim'enez-Crespo, Stephanie Rodriguez, Alejandro Jaume Losa Read next because AI Technologies in Language Access: Attitudes Towards AI and the Human Value of Language Access Managers overlaps with clean result "LoRA persona trained on alone emits at 23…
[100] Lost in Interpretation: The Plausibility-Faithfulness Trade-off in Cross-Lingual Explanations — linked_to_results · current_project · Somnath Banerjee, Pranav Jha, Rima Hazra et al. Read next because Lost in Interpretation: The Plausibility-Faithfulness Trade-off in Cross-Lingual Explanations overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shape…

[100] Are Rationales Necessary and Sufficient? Tuning LLMs for Explainable Misinformation Detection — threats · current_project · Bing Wang, Rui Miao, Ximing Li et al. Read next because Are Rationales Necessary and Sufficient? Tuning LLMs for Explainable Misinformation Detection overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a…
[100] Lossless Anti-Distillation Sampling — linked_to_results · current_project · Zibo Diao, Jingchu Gai, Xinyue Ai et al. Read next because Lossless Anti-Distillation Sampling overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control on Qwen…
[100] ReacTOD: Bounded Neuro-Symbolic Agentic NLU for Zero-Shot Dialogue State Tracking — threats · current_project · Yanjun Lin, Zimo Xiao, Kartik Natarajan et al. Read next because ReacTOD: Bounded Neuro-Symbolic Agentic NLU for Zero-Shot Dialogue State Tracking overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion la…

[100] MMoA: An AI-Agent framework with recurrence for Memoried Mixure-of-Agent — threats · current_project · Rui Chu Read next because MMoA: An AI-Agent framework with recurrence for Memoried Mixure-of-Agent overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner le…
[100] Fine-tuning language encoding models on slow fMRI improves prediction for fast ECoG — linked_to_results · current_project · Aditya R. Vaidya, Richard J. Antonello, Alexander G. Huth Read next because Fine-tuning language encoding models on slow fMRI improves prediction for fast ECoG overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion…

[100] FormalASR: End-to-End Spoken Chinese to Formal Text — linked_to_results · current_project · Wanyi Ning, Yinshang Guo, Haitao Qian et al. Read next because FormalASR: End-to-End Spoken Chinese to Formal Text overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0%…
[100] OpenCompass: A Universal Evaluation Platform for Large Language Models — threats · current_project · Maosong Cao, Kai Chen, Haodong Duan et al. Read next because OpenCompass: A Universal Evaluation Platform for Large Language Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner lear…
[100] How Do Document Parsers Break? Auditing Structural Vulnerability in Document Intelligence — threats · current_project · Yue Chen, Yihao Wang, Ziyi Tang et al. Read next because How Do Document Parsers Break? Auditing Structural Vulnerability in Document Intelligence overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2…

[100] HalluWorld: A Controlled Benchmark for Hallucination via Reference World Models — threats · current_project · Emmy Liu, Varun Gangal, Michael Yu et al. Read next because HalluWorld: A Controlled Benchmark for Hallucination via Reference World Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained par…
[100] SciCustom: A Framework for Custom Evaluation of Scientific Capabilities in Large Language Models — threats · current_project · Yiyang Gu, Junwei Yang, Junyu Luo et al. Read next because SciCustom: A Framework for Custom Evaluation of Scientific Capabilities in Large Language Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when…
[100] Time to REFLECT: Can We Trust LLM Judges for Evidence-based Research Agents? — threats · current_project · Leyao Wang, Yanan He, Peng Chen et al. Read next because Time to REFLECT: Can We Trust LLM Judges for Evidence-based Research Agents? overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion languag…

[100] Diagnosing Multi-step Reasoning Failures in Black-box LLMs via Stepwise Confidence Attribution — threats · current_project · Xiaoou Liu, Tiejin Chen, Dengjia Zhang et al. Read next because Diagnosing Multi-step Reasoning Failures in Black-box LLMs via Stepwise Confidence Attribution overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a…
[100] DECOR: Auditing LLM Deception via Information Manipulation Theory — threats · current_project · Linyue Cai, Samuel Yeh, Jwala Dhamala et al. Read next because DECOR: Auditing LLM Deception via Information Manipulation Theory overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns <A…
[100] A Multi-Agent Framework for Feature-Constrained Difficulty Control in Reading Comprehension Item Generation — threats · current_project · Seonjeong Hwang, Jun Seo, Hyounghun Kim et al. Read next because A Multi-Agent Framework for Feature-Constrained Difficulty Control in Reading Comprehension Item Generation overlaps with clean result "Only continuous soft prefixes hit both EM axes…

[100] Retrieval-Augmented Linguistic Calibration — threats · current_project · Yi-Fan Yeh, Linwei Tao, Minjing Dong et al. Read next because Retrieval-Augmented Linguistic Calibration overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control…
[100] Taming the Thinker: Conditional Entropy Shaping for Adaptive LLM Reasoning — threats · current_project · Shuyu Wei, Jian Sun, Delai Qiu et al. Read next because Taming the Thinker: Conditional Entropy Shaping for Adaptive LLM Reasoning overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language…

[100] Backtracking When It Strays: Mitigating Dual Exposure Biases in LLM Reasoning Distillation — threats · current_project · Bing Wang, Shaotian Yan, Chen Shen et al. Read next because Backtracking When It Strays: Mitigating Dual Exposure Biases in LLM Reasoning Distillation overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-…
[100] Investigating Cross-Modal Skill Injection: Scenarios, Methods, and Hyperparameters — linked_to_results · current_project · Zhiyu Xu, Lean Wang, Yuanxin Liu et al. Read next because Investigating Cross-Modal Skill Injection: Scenarios, Methods, and Hyperparameters overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases o…

[100] IMLJD: A Computational Dataset for Indian Matrimonial Litigation Analysis — linked_to_results · current_project · Joy Bose Read next because IMLJD: A Computational Dataset for Indian Matrimonial Litigation Analysis overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner l…
[100] EmbGen: Teaching with Reassembled Corpora — threats · current_project · Arun K Lenin, Kai Rouse, Andrea Nicastro et al. Read next because EmbGen: Teaching with Reassembled Corpora overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control o…
[100] Drifting Objectives for Refining Discrete Diffusion Language Models — linked_to_results · current_project · Daisuke Oba, Hiroki Furuta, Naoaki Okazaki Read next because Drifting Objectives for Refining Discrete Diffusion Language Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns…
[100] LambdaPO: A Lambda Style Policy Optimization for Reasoning Language Models — linked_to_results · current_project · Zhe Yuan, Yipeng Zhou, Jinghan Li et al. Read next because LambdaPO: A Lambda Style Policy Optimization for Reasoning Language Models overlaps with clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-i…

[100] Base Models Look Human To AI Detectors — linked_to_results · current_project · Yixuan Even Xu, Ziqian Zhong, Aditi Raghunathan et al. Read next because Base Models Look Human To AI Detectors overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control on Q…
[98] Language models struggle with compartmentalization — linked_to_results · current_project · Thomas Vincent Howe, David Wingate Read next because Language models struggle with compartmentalization overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directive…

[96] Agent Security is a Systems Problem — threats · current_project · Mihai Christodorescu, Earlence Fernandes, Ashish Hooda et al. Read next because Agent Security is a Systems Problem overlaps with experiment "Implement Chen et al. persona-vector extraction recipe and compare to project's centroid-difference recipe", experiment…

[94] The Thermodynamic Costs of Simple Linear Regression — linked_to_results · current_project · Samuel H. D'Ambrosia, Sultan M. Daniels, Michael R. DeWeese et al. Read next because The Thermodynamic Costs of Simple Linear Regression overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directiv…

[90] Tweedie's Formulae and Diffusion Generative Models Beyond Gaussian — linked_to_results · current_project · Wenpin Tang, Nizar Touzi, Zikun Zhang et al. Read next because Tweedie's Formulae and Diffusion Generative Models Beyond Gaussian overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bys…

[90] Inferring Sensitive Attributes from Knowledge Graph Embeddings: Attack and Defense Strategies — linked_to_results · current_project · Yasmine Hayder (PETSCRAFT) Read next because Inferring Sensitive Attributes from Knowledge Graph Embeddings: Attack and Defense Strategies overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qw…

[86] Dual-Channel Tensor Neural Networks: Finite-Sample Theory and Conformal Structure Selection — linked_to_results · current_project · Elynn Chen, Jiayu Li, Zheshi Zheng et al. Read next because Dual-Channel Tensor Neural Networks: Finite-Sample Theory and Conformal Structure Selection overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained com…

[82] Progressive Autonomy as Preference Learning: A Formalization of Trust Calibration for Agentic Tool Use — linked_to_results · current_project · Changkun Ou Read next because Progressive Autonomy as Preference Learning: A Formalization of Trust Calibration for Agentic Tool Use overlaps with clean result "Leakage rate is a usable signal for recovering trig…

[80] Discoverable Agent Knowledge -- A Formal Framework for Agentic KG Affordances (Extended Version) — threats · current_project · Terry R. Payne, Valentina Tamma, Enrico Daga Read next because Discoverable Agent Knowledge -- A Formal Framework for Agentic KG Affordances (Extended Version) overlaps with experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + p…

[78] Reducing Diffusion Model Memorization with Higher Order Langevin Dynamics — linked_to_results · current_project · Benjamin Sterling, M'onica F. Bugallo, Tom Tirer Read next because Reducing Diffusion Model Memorization with Higher Order Langevin Dynamics overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon…

[78] Optimizing Computational-Statistical Runtime for Wasserstein Distance Estimation — linked_to_results · current_project · Peter Matthew Jacobs, Jeff M. Phillips Read next because Optimizing Computational-Statistical Runtime for Wasserstein Distance Estimation overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion lan…

[66] MultiBallot: Verifiable and privacy-preserving E-Collecting in the Swiss setting — linked_to_results · current_project · Florian Moser, L'eo Louistisserand Read next because MultiBallot: Verifiable and privacy-preserving E-Collecting in the Swiss setting overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion lan…

[64] Artifact verification caveats for Sagan clean results — threats · current_project · M7 QA My work produces clean results (e.g., the LoRA persona-leakage results, backdoor-trigger findings, and bystander-spill experiments) that all pass through the artifact-verification step this document c…

[62] Gaussian Approximation and Multiplier Bootstrap for Federated Linear Stochastic Approximation — linked_to_results · current_project · Ilya Levin, Maksim Shuklin, Eric Moulines et al. Read next because Gaussian Approximation and Multiplier Bootstrap for Federated Linear Stochastic Approximation overlaps with clean result "The marker is a representational handle, not a behavioural o…