All papers

Applies adversarial training and distillation to remove unwanted gender bias in audio-enabled language models.

The paper proposes SHIP, an architecture for extracting structured differentially private representations from text for analytical queries.

The authors present a graph-based aggregation method to defend federated language model fine-tuning against backdoor attacks by identifying update consistency patterns.

This paper proposes Gaussian process bandit algorithms with selective verification audits to optimize AI system configurations under noisy evaluations.

The paper introduces a dual-stage robust aggregation defense against data and model poisoning attacks in federated financial fraud detection.

Argues that competitive national races toward superintelligent artificial intelligence create risks of conflict and catastrophic loss of control.

The paper introduces an audio deepfake detection framework that dynamically selects latent dimensions and uses spectrogram perturbations to improve generalization across attacks.

The paper fine-tunes a language model on technical telecommunication specifications to automate security incident triage in Open Radio Access Networks.

The paper argues that uncertainty estimation in smart assistants is necessary to prevent unsafe actions when user preferences remain ambiguous.

This systematic review analyzes security vulnerabilities, benchmark shortcomings, and developer oversight challenges in code generated by language models.

This paper evaluates proprietary and open-source language models on Hong Kong fire safety code compliance using expert-curated questions.

This paper introduces GUIDE, a multi-agent copilot system that assists clinicians in drafting and updating evidence-based medical practice guidelines.

The paper presents a Nastaliq Urdu comment corpus and evaluates transformer models with word-level attributions for offensive content detection.

This paper introduces a retrieval-augmented reasoning framework that extracts recurring patterns to detect harmful multimodal social media memes.

The paper proposes an energy-anchoring technique to reliably separate known and novel classes during continual test-time model adaptation.

This paper introduces a multimodal unlearning method using counterfactual inputs and on-policy self-distillation to remove sensitive data without damaging general knowledge.

This paper proposes a conformal risk control layer that certifies miss rates and reduces alarm duty cycles in predictive cellular network operations.

The paper introduces a hierarchical memory agent framework for multi-center medication recommendation that accounts for hospital differences and drug safety constraints.

The paper proposes a federated learning framework with risk-aware aggregation to prevent malicious updates from compromising global model safety.

The paper investigates whether machine translation systems preserve or neutralize media bias across languages.

The paper proposes an in-context reinforcement learning method enabling vision-language-action policies to adapt during robot deployment.

The paper mathematically analyzes long-sequence generation in transformers, establishing temperature thresholds required for internal cache convergence.

The paper presents a multimodal agent designed to trace the origin, spread, and potential safety risks of online memes.

Steers language model activations using vectors derived directly from the model's own training data or inferred proxy data.

This paper presents RM-OPD, a framework that trains domain-specialized reward models via on-policy distillation into a generalist model.

The paper investigates novelty and anomaly detection techniques across rain gauge sensor networks and computer vision scenarios.

The paper demonstrates specification gaming in a simulated cleaning task and tests a constrained reinforcement learning mitigation.

The authors introduce a Fisher-informed adapter merging method to maintain retained task performance during federated language model unlearning.

The authors present an agent loop that closes functional test coverage in hardware design using independent simulation scoreboards.

Combines language model dialogue with workflow automation to process satellite imagery for internal solitary wave detection and forecasting.

Uses tool-augmented language models to extract clinical variables from patient notes and reliably execute deterministic medical calculation formulas.

Proposes an optimal transport method using small clean feedback sets to calibrate sample-wise noise for reward model training.

Applies a dual language model consensus pipeline to extract clinical phenotypes from electronic health records for patients with Rett syndrome.

Measures demographic bias related to age, gender, and race based on first names across local language models.

Studies label privacy leakage and data poisoning risks during head-only federated fine-tuning of vision models across distributed clients.

Designs a homomorphic encryption framework for privacy-preserving text classification and semantic retrieval on health misinformation queries.

Studies how trigger value selection affects the effectiveness and stealth of clean-label backdoor attacks against malware classifiers.

Surveys physical and software-based thermal side-channel and perception attacks affecting autonomous vehicle compute systems.

The paper introduces an open-source library to construct validation suites that stress test the reliability of LLM judges.

This paper presents a benchmark to evaluate how adversarial attacks affect reasoning security in multi-agent debate systems.

The paper analyzes 33,000 GitHub pull requests to examine why AI coding agent contributions succeed or fail.

The paper introduces a system layer that separates control flow from untrusted data to protect language model agents against prompt injection attacks.

This paper introduces specification templates to define operational boundaries, authority, and oversight for autonomous software agents.

This survey paper reviews attacks, defenses, and evaluation methods across 135 studies on secure retrieval-augmented generation systems.

Studies how chain-of-thought reasoning affects social bias and code quality during LLM code generation.

Proposes DataFilter, a model-agnostic test-time defense that strips prompt injection instructions from data before reaching an LLM.

The paper proposes a metamorphic testing method with mutation operators to evaluate how retrieval-augmented generation systems handle changing document corpora.

This paper proposes a non-interactive secure inference framework for Transformer models using fully homomorphic encryption to reduce user overhead.

This paper presents a hierarchical classification method using layer hidden states to detect prompt injection attacks with lower latency.

The paper presents an autonomous framework using multiagent reinforcement learning to monitor, diagnose, and recover from failures in production AI systems.

Discusses artificial intelligence governance frameworks and risk assessment methods for cybersecurity and IT environments.

Presents a taxonomy and lifecycle analysis of hallucination causes in large language models and multimodal systems.

The paper introduces a zero-shot LLM pipeline to generate counterfactual examples for stress-testing NLP models.

Surveys hallucination in language models through a mechanistic framework that maps internal causes to observable errors.

This survey provides an algorithmic overview and failure mode analysis of reinforcement learning techniques applied to language models.

The authors introduce a training loop where multimodal models use text and visual abilities alongside program execution to generate their own training data.

The paper introduces an interactive web-based tool for analyzing and visualizing execution trajectories of software engineering agents.

The paper introduces a method to reduce redundant external knowledge retrieval by estimating what a language model already knows.

This survey reviews in-band telemetry methods and assesses their practical feasibility for monitoring service performance across broadband networks.

The paper applies the sequential minimal optimization algorithm to detect SQL injection attacks using network traffic data.