11 papers
Self-Healing Harness for Runtime Oversight of Agent Self-Modification
Sina Tayebati, Divake Kumar, Nastaran Darabi +2
LLM agents can change their own future behavior, raising a basic control question of which self-generated changes should be allowed to persist. We formulate this as admission contr…
CURA: Certified Runtime Alarms for Computer-Use Agents
Divake Kumar, Sina Tayebati, Devashri Naik +5
Self-report is the cheapest oversight channel a deployer has, and on capable computer-use agents (CUAs) it fails precisely where oversight matters. On 361 OSWorld tasks our pipelin…
Uncertainty Quantification for Computer-Use Agents: A Benchmark across Vision-Language Models and GUI Grounding Datasets
Divake Kumar, Sina Tayebati, Devashri Naik +5
Computer-use agents turn vision-language model (VLM) predictions into executable GUI clicks, so reliable uncertainty estimates are essential for rejection, calibration, miss-severi…
VLM Judges Can Rank but Cannot Score: Task-Dependent Uncertainty in Multimodal Evaluation
Divake Kumar, Sina Tayebati, Devashri Naik +2
Vision-language models (VLMs) are increasingly used as automated judges for multimodal systems, yet their scores provide no indication of reliability. We study this problem through…
TRACER: Trajectory Risk Aggregation for Critical Episodes in Agentic Reasoning
Sina Tayebati, Divake Kumar, Nastaran Darabi +3
Estimating uncertainty for AI agents in real-world multi-turn tool-using interaction with humans is difficult because failures are often triggered by sparse critical episodes (e.g.…
Learnable Conformal Prediction with Context-Aware Nonconformity Functions for Robotic Planning and Perception
Divake Kumar, Sina Tayebati, Francesco Migliarba +2
Deep learning models in robotics often output point estimates with poorly calibrated confidences, offering no native mechanism to quantify predictive reliability under novel, noisy…
EigenTrack: Spectral Activation Feature Tracking for Hallucination and Out-of-Distribution Detection in LLMs and VLMs
Davide Ettori, Nastaran Darabi, Sina Tayebati +4
Large language models (LLMs) offer broad utility but remain prone to hallucination and out-of-distribution (OOD) errors. We propose EigenTrack, an interpretable real-time detector…
EigenShield: Causal Subspace Filtering via Random Matrix Theory for Adversarially Robust Vision-Language Models
Nastaran Darabi, Devashri Naik, Sina Tayebati +3
Vision-Language Models (VLMs) inherit adversarial vulnerabilities of Large Language Models (LLMs), which are further exacerbated by their multimodal nature. Existing defenses, incl…
Learning Conformal Abstention Policies for Adaptive Risk Management in Large Language and Vision-Language Models
Sina Tayebati, Divake Kumar, Nastaran Darabi +3
Large Language and Vision-Language Models (LLMs/VLMs) are increasingly used in safety-critical applications, yet their opaque decision-making complicates risk assessment and reliab…
SPARC: Subspace-Aware Prompt Adaptation for Robust Continual Learning in LLMs
Dinithi Jayasuriya, Sina Tayebati, Davide Ettori +2
We propose SPARC, a lightweight continual learning framework for large language models (LLMs) that enables efficient task adaptation through prompt tuning in a lower-dimensional sp…
Enhancing Trust in Large Language Models via Uncertainty-Calibrated Fine-Tuning
Ranganath Krishnan, Piyush Khanna, Omesh Tickoo
Large language models (LLMs) have revolutionized the field of natural language processing with their impressive reasoning and question-answering capabilities. However, these models…