Nastaran Darabi — accepted AI-conference papers

2026

EigenShield: Inference-Time, Model-Agnostic Jailbreaking Defense via Causal Subspace Filtering

Nastaran Darabi, Devashri Naik, Sina Tayebati, Dinithi Jayasuriya, Ranganath Krishnan, Amit Ranjan Trivedi

AAAI 2026technical

Large Language Models (LLMs) and Vision-Language Models (VLMs) remain highly vulnerable to adversarial attacks despite widespread adoption. Existing defenses typically require retraining, rely on heuristics, or fail under adaptive and out-of-distribution (OOD) conditions. We introduce EigenShield, a

Cited by 0 Source ↗PDF ↗Scholar ↗

2026

INTACT: INDUCING NOISE TOLERANCE THROUGH ADVERSARIAL CURRICULUM TRAINING FOR LIDAR-BASED SAFETY-CRITICAL PERCEPTION AND AUTONOMY

Nastaran Darabi

ICASSP 2026oral

In this work, we present INTACT, a novel two-phase framework designed to enhance the robustness of deep neural networks (DNNs) against noisy LiDAR data in safety-critical perception tasks. INTACT combines meta-learning with adversarial curriculum training (ACT) to systematically address challenges p…

Cited by 0 Source ↗PDF ↗Scholar ↗

2026

Resilience in Ambient Multi-Agent LLMs via Decentralized Bio-Autonomic Control and Immune-Inspired Anomaly Detection

Nastaran Darabi, Devashri Naik, Sina Tayebati, Dinithi Jayasuriya, Amit Ranjan Trivedi

AAAI 2026technical

Large Language Model (LLM) agents are now widely deployed in Ambient Intelligence (AmI) environments, where autonomous agents must sense, act, and coordinate at scale. As agent capabilities and interdependence increase, traditional reliability strategies such as isolated adaptive control, anomaly de

Cited by 0 Source ↗PDF ↗Scholar ↗

2026

TRACER: Trajectory Risk Aggregation for Critical Episodes in Agentic Reasoning

Sina Tayebati, Divake Kumar, Nastaran Darabi, Davide Ettori, Ranganath Krishnan, Amit Trivedi

ICML 2026poster

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., looping, incoherent tool use, or user-agent miscoordination) even when local generation appears confident. Existing un…

Cited by 0 Source ↗Scholar ↗

2025

Enhancing 3D Robotic Vision Robustness by Minimizing Adversarial Mutual Information through Curriculum Training

Nastaran Darabi, Dinithi Jayasuriya, Devashri Naik, Theja Tulabandhula, Amit Ranjan Trivedi

ICRA 2025

Adversarial attacks exploit vulnerabilities in a model's decision boundaries through small, carefully crafted perturbations that lead to significant mispredictions. In 3D vision, the high dimensionality and sparsity of data greatly expand the attack surface, making 3D vision particularly vulnerable

Cited by 2 Source ↗code ↗Scholar ↗

2024

Conformalized Multimodal Uncertainty Regression and Reasoning

Domenico Parente, Nastaran Darabi, Alex C. Stutts, Theja Tulabandhula, Amit Ranjan Trivedi

ICASSP 2024accepted

This paper introduces a lightweight uncertainty estimator capable of predicting multimodal (disjoint) uncertainty bounds by integrating conformal prediction with a deep-learning regressor. We specifically discuss its application for visual odometry (VO), where environmental features such as flying d…

Cited by 0 Source ↗Scholar ↗