Image de couverture de livre

Microsaccade-Inspired Probing: Positional Encoding Perturbations Reveal LLM Misbehaviours

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Melo, Rui, Abreu, Rui, Pasareanu, Corina S.
Format:	Preprint
Publié:	2025
Sujets:	Machine Learning Artificial Intelligence
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Concept-Based Abductive and Contrastive Explanations for Behaviors of Vision Models
par: Canizales, Ronaldo, et autres
Publié: (2026)

Debugging and Runtime Analysis of Neural Networks with VLMs (A Case Study)
par: Hu, Boyue Caroline, et autres
Publié: (2025)

Algebraic Positional Encodings
par: Kogkalidis, Konstantinos, et autres
Publié: (2023)

Are Sparse Autoencoders Useful for Java Function Bug Detection?
par: Melo, Rui, et autres
Publié: (2025)

Rotary Position Encodings for Graphs
par: Reid, Isaac, et autres
Publié: (2025)

SYNAPSE: Framework for Neuron Analysis and Perturbation in Sequence Encoding
par: Ochoa, Jesús Sánchez, et autres
Publié: (2026)

Conformal Safety Shielding for Imperfect-Perception Agents
par: Scarbro, William, et autres
Publié: (2025)

On the Stability of Expressive Positional Encodings for Graphs
par: Huang, Yinan, et autres
Publié: (2023)

Graph Transformers without Positional Encodings
par: Garg, Ayush
Publié: (2024)

Benchmarking Positional Encodings for GNNs and Graph Transformers
par: Grötschla, Florian, et autres
Publié: (2024)

APEX: Probing Neural Networks via Activation Perturbation
par: Ren, Tao, et autres
Publié: (2026)

Understanding and Improving Laplacian Positional Encodings For Temporal GNNs
par: Galron, Yaniv, et autres
Publié: (2025)

Learnable Spatial-Temporal Positional Encoding for Link Prediction
par: Tieu, Katherine, et autres
Publié: (2025)

CoPE: A Lightweight Complex Positional Encoding
par: Amballa, Avinash
Publié: (2025)

Group Representational Position Encoding
par: Zhang, Yifan, et autres
Publié: (2025)

Probing the Embedding Space of Transformers via Minimal Token Perturbations
par: Conti, Eddie, et autres
Publié: (2025)

Graph Self-Supervised Learning with Learnable Structural and Positional Encodings
par: Wijesinghe, Asiri, et autres
Publié: (2025)

Graph Alignment for Benchmarking Graph Neural Networks and Learning Positional Encodings
par: Lagesse, Adrien, et autres
Publié: (2025)

GUI-Perturbed: Domain Randomization Reveals Systematic Brittleness in GUI Grounding Models
par: Wang, Yangyue, et autres
Publié: (2026)

What Makes Quantization for Large Language Models Hard? An Empirical Study from the Lens of Perturbation
par: Gong, Zhuocheng, et autres
Publié: (2024)

MEP: Multiple Kernel Learning Enhancing Relative Positional Encoding Length Extrapolation
par: Gao, Weiguo
Publié: (2024)

Structural Positional Encoding for knowledge integration in transformer-based medical process monitoring
par: Irwin, Christopher, et autres
Publié: (2024)

Mesa-Extrapolation: A Weave Position Encoding Method for Enhanced Extrapolation in LLMs
par: Ma, Xin, et autres
Publié: (2024)

Fighting AI with AI: Leveraging Foundation Models for Assuring AI-Enabled Safety-Critical Systems
par: Mavridou, Anastasia, et autres
Publié: (2025)

Joint Encoding of KV-Cache Blocks for Scalable LLM Serving
par: Kampeas, Joseph, et autres
Publié: (2026)

Epi$^2$-Net: Advancing Epidemic Dynamics Forecasting with Physics-Inspired Neural Networks
par: Sun, Rui, et autres
Publié: (2025)

Beyond KL Divergence: Policy Optimization with Flexible Bregman Divergences for LLM Reasoning
par: Yuan, Rui, et autres
Publié: (2026)

Boosting LLM Reasoning via Human-Inspired Reward Shaping
par: Lin, Wenze, et autres
Publié: (2026)

Hessian Aware Low-Rank Perturbation for Order-Robust Continual Learning
par: Li, Jiaqi, et autres
Publié: (2023)

Position: Let's Develop Data Probes to Fundamentally Understand How Data Affects LLM Performance
par: Wang, Shiqiang, et autres
Publié: (2026)

Adaptive Layerwise Perturbation: Unifying Off-Policy Corrections for LLM RL
par: Ye, Chenlu, et autres
Publié: (2026)

Plausibility Is Not Prediction: Contrastive Evidence for LLM-Based Cellular Perturbation Reasoning
par: Yuan, Xinyu, et autres
Publié: (2026)

Revealing Combinatorial Reasoning of GNNs via Graph Concept Bottleneck Layer
par: Niu, Yue, et autres
Publié: (2026)

OPPO: Bayesian Value Recursion for Token-Level Credit Assignment in LLM Reasoning
par: Li, Yu, et autres
Publié: (2026)

LLM-Inspired Pretrain-Then-Finetune for Small-Data, Large-Scale Optimization
par: Zhang, Zishi, et autres
Publié: (2026)

Perturbation-efficient Zeroth-order Optimization for Hardware-friendly On-device Training
par: Tan, Qitao, et autres
Publié: (2025)

Graph Federated Learning Based Proactive Content Caching in Edge Computing
par: Wang, Rui
Publié: (2025)

Of All StrIPEs: Investigating Structure-informed Positional Encoding for Efficient Music Generation
par: Agarwal, Manvi, et autres
Publié: (2025)

A Small Math Model: Recasting Strategy Choice Theory in an LLM-Inspired Architecture
par: Rahman, Roussel, et autres
Publié: (2025)

Adaptive Repetition for Mitigating Position Bias in LLM-Based Ranking
par: Vardasbi, Ali, et autres
Publié: (2025)