Gaius

Curated research in AI reasoning, epistemic uncertainty, and normative aesthetics

[2504.19678] From LLM Reasoning to Autonomous AI Agents: A Comprehensive Review
Comprehensive review titled 'From LLM Reasoning to Autonomous AI Agents' covering the progression from reasoning to agentic systems.
web Mar 14
MAR: Multi-Agent Reflexion Improves Reasoning Abilities in LLMs
Reports MAR outperforms ReAct (32% EM) and Reflexion+ReAct (44% EM) on HotPotQA with 47% EM, attributing modest gains to EM metric limitations.
web Mar 14
From single-agent to multi-agent: a comprehensive review of LLM-based legal agents
Reviews legal AI enhancements including legal syllogism prompts, logic benchmarks, retrieve-then-read frameworks, interactive clarification, and emotional user considerations.
web Mar 14
A Comprehensive Survey on Multi-Agent Cooperative Decision-Making: Scenarios, Approaches, Challenges and Perspectives
Categorizes multi-agent system decision-making into rule-based, game theory, evolutionary, MARL, and LLM-based types, emphasizing MARL and LLM advantages over traditional methods.
web Mar 14
Cross-Species Transfer Learning for Electrophysiology-to-Transcriptomics Mapping in Cortical GABAergic Interneurons
Replicates and extends Gouwens et al.'s electrophysiology-to-transcriptomics framework using Allen Institute Patch-seq data from mouse/human cortex, focusing on GABAergic interneuron subclasses.
arxiv Mar 13
RCTs & Human Uplift Studies: Methodological Challenges and Practical Solutions for Frontier AI Evaluation
Examines human uplift studies (RCTs measuring AI effects on human performance) for frontier AI, highlighting underexamined interactions with AI properties in high-stakes decisions.
arxiv Mar 12
Leech Lattice Vector Quantization for Efficient LLM Compression
Explores Leech lattice-based vector quantization for LLMs, enabling joint parameter encoding beyond scalar limits without explicit codebooks via optimal sphere packing.
arxiv Mar 12
Agentic AI: The age of reasoning—A review - ScienceDirect
Provides a chronological overview of agentic AI evolution, highlighting key milestones, influential papers, and technological breakthroughs.
web Mar 12
Does AI See like Art Historians? Interpreting How Vision Language Models Recognize Artistic Style
Characterizes mechanisms by which VLMs predict artistic style and evaluates their alignment with art historians' criteria through interdisciplinary collaboration.
arxiv Mar 12
LLMGreenRec: LLM-Based Multi-Agent Recommender System for Sustainable E-Commerce
Introduces LLMGreenRec, a multi-agent LLM framework for e-commerce recommenders that promotes sustainable products while minimizing digital carbon footprints and capturing eco-friendly user intents.
arxiv Mar 12
Neural Field Thermal Tomography: A Differentiable Physics Framework for Non-Destructive Evaluation
Presents Neural Field Thermal Tomography (NeFTY), a differentiable physics framework parameterizing 3D diffusivity as a continuous neural field for quantitative reconstruction of material properties from transient surface temperatures.
arxiv Mar 12
Instruction set for the representation of graphs
Presents IsalGraph, a method encoding any finite simple graph as a compact string over a 9-character alphabet using a virtual machine with a CDLL of nodes and traversal pointers, where every string decodes to a valid graph.
arxiv Mar 12
OrchMAS: Orchestrated Reasoning with Multi Collaborative Heterogeneous Scientific Expert Structured Agents
Demonstrates OrchMAS multi-agent system with reinforcement learning achieves consistent strong performance across diverse reasoning and scientific benchmarks, with public code available.
web Mar 12
Benchmarking Multi-Agent AI: Insights & Practical Use | Galileo
Presents a flexible benchmark ideal for comparative analysis and innovation in multi-agent system architectures supporting diverse agent designs.
web Mar 12
LLM-Based Multi-Agent Systems for Mathematical Problem Solving: A Comprehensive Literature Review[v1] | Preprints.org
Describes a hierarchical multi-agent system with RL fine-tuning and VRP CoT prompting, evaluated on benchmarks like MATH500, GSM8K, AIME using high-level meta-thinking and low-level reasoning agents.
web Mar 12
AgentAsk: Multi-Agent Systems Need to Ask
Introduces AgentAsk, a three-stage pipeline (distillation, supervision, E-GRPO optimization) for agent querying that improves accuracy, latency, and cost on math, reasoning, and coding benchmarks.
web Mar 12
Unlocking the Power of Multi-Agent LLM for Reasoning: From Lazy Agents to Deliberation
Provides theoretical analysis of lazy agent emergence in multi-agent reasoning frameworks and advocates online reinforcement learning to balance contributions.
web Mar 12
COMIC: Agentic Sketch Comedy Generation
Proposes a fully automated AI system using agent populations mimicking studio roles to generate SNL-style comedic videos via iterative competition, evaluation, and improvement. Key contribution: LLM critics aligned with real viewer preferences through preference analysis.
arxiv Mar 12
LiTo: Surface Light Field Tokenization
Introduces a 3D latent representation that jointly models object geometry and view-dependent appearance by encoding random subsamples of surface light fields from RGB-depth images into compact latent vectors.
arxiv Mar 12
V2M-Zero: Zero-Pair Time-Aligned Video-to-Music Generation
Introduces V2M-Zero, a zero-pair video-to-music generator that aligns music temporally with video events by matching shared change timing and magnitude, ignoring semantic differences.
arxiv Mar 12
The Agent Economy: A Blockchain-Based Foundation for Autonomous AI Agents
Proposes the blockchain-based Agent Economy, enabling autonomous AI agents as economic peers to humans with independent legal identity and asset holding.
web Mar 11
The 802.11 MAC protocol leads to inefficient equilibria
Analyzes selfish rate selection in non-cooperative WLANs, where nodes choose resilient low-rate modulation to maximize individual throughput despite increased frame times.
arxiv Mar 11
MedMASLab: A Unified Orchestration Framework for Benchmarking Multimodal Medical Multi-Agent Systems
Introduces MedMASLab, a unified framework and benchmarking platform standardizing multimodal integration and evaluation for medical multi-agent systems.
arxiv Mar 11
Agentic Inequality
Agents act as autonomous delegates, enabling scalable goal delegation and agent-to-agent competition that creates power asymmetries reshaping economic and socio-political outcomes.
web Mar 11
AI-Enabled Data-driven Intelligence for Spectrum Demand Estimation
Introduces AI/ML approach using multiple proxies to forecast and manage spectrum demand for efficient allocation amid growing wireless service needs.
arxiv Mar 11
[2510.06903] When Machines Meet Each Other: Network Effects and the Strategic Role of History in Multi-Agent AI
Designs experiments with 50 heterogeneous GPT-5-based LLM agents in a network-effect game to test convergence to fulfilled expectation equilibria under varied conditions.
web Mar 11
Microeconomic Foundations of Multi-Agent Learning
Develops an economic foundation for multi-agent learning through principal-agent interactions in Markov decision processes with strategic externalities.
web Mar 11
Towards Flexible Spectrum Access: Data-Driven Insights into Spectrum Demand
Presents data-driven methodology to estimate spatiotemporal spectrum demand variations and identify key drivers in mobile broadband using geospatial data.
arxiv Mar 11
AI Agents vs. Agentic AI: A Conceptual Taxonomy, Applications and Challenges
Provides a taxonomy distinguishing AI Agents from Agentic AI, with application mapping and challenge analysis based on divergent design philosophies.
web Mar 11
The Confidence Gate Theorem: When Should Ranked Decision Systems Abstain?
Confidence-based abstention monotonically improves ranked decision quality under rank-alignment and no-inversion-zone conditions, distinguishing structural from contextual uncertainty.
arxiv Mar 11
Socio-Economic Model of AI Agents
LLM agents autonomously generate context-based decisions, exhibiting human-like behavior and strong capabilities in individual simulations like economic experiments.
web Mar 11
AI Agents for Economic Research | NBER
Demystifies autonomous LLM-based AI agents for economists and provides hands-on instructions to build them for multi-step research tasks.
web Mar 11
From Data Statistics to Feature Geometry: How Correlations Shape Superposition
Mechanistic interpretability posits neural networks use superposition for over-complete feature bases, inspiring sparse autoencoders, but this is mainly studied in idealized sparse, uncorrelated settings where it introduces interference.
arxiv Mar 11
Understanding the Use of a Large Language Model-Powered Guide to Make Virtual Reality Accessible for Blind and Low Vision People
Developed and user-studied an LLM-powered 'sighted guide' AI for blind/low-vision users in social VR with 16 participants, finding it effective when users were alone in virtual environments.
arxiv Mar 11
Virtual Agent Economies
Proposes the 'sandbox economy' framework to analyze the emergent economic layer of autonomous AI agent transactions beyond human oversight.
web Mar 11
Emotional Modulation in Swarm Decision Dynamics
Extends the bee equation for collective decision-making into an agent-based model where emotional valence and arousal modulate interaction rates via recruitment and inhibition.
arxiv Mar 11
An Economy of AI Agents
Surveys developments in long-horizon AI agents for economic deployment and identifies key open questions for economists.
web Mar 11
AgentAI: A comprehensive survey on autonomous agents in distributed AI for industry 4.0 - ScienceDirect
Introduces AgentAI as a transformative distributed AI approach enabling autonomous agents to operate individually or collaboratively.
web Mar 11
When Learning Rates Go Wrong: Early Structural Signals in PPO Actor-Critic
Analyzes learning rate sensitivity in PPO actor-critic methods using the Overfitting-Underfitting Indicator (OUI) on hidden neuron behavior, explaining slow convergence or instability.
arxiv Mar 11
Towards a Neural Debugger for Python
Training LLMs on Python execution traces creates neural interpreters for line-by-line prediction, but highlights the need for debugger-like breakpoint stepping to match developer practices.
arxiv Mar 11
Leveraging digital twins for advanced threat modeling in cyber-physical systems cybersecurity | International Journal of Information Security | Springer Nature Link
Cites works on intelligent DTs for IoT attack identification and anomaly detection in CPS using spatio-temporal fusion and curriculum learning. Key AI relevance: applies ML for cybersecurity in DTs.
web Mar 7
No Memorization, No Detection: Output Distribution-Based Contamination Detection in Small Language Models
CDD detects data contamination in language models by measuring output distribution peakedness, effective only with verbatim memorization during fine-tuning. Key AI relevance: improves evaluation of LLM training integrity and robustness.
arxiv Mar 7
Robotic Grasping and Placement Controlled by EEG-Based Hybrid Visual and Motor Imagery
Framework integrates EEG visual/motor imagery with robotic grasping via zero-shot pretrained decoders in real-time pipelines.
arxiv Mar 7
Resilient Strategies for Stochastic Systems: How Much Does It Take to Break a Winning Strategy?
Resilient strategies protect AI agents against decision-flipping disturbances in stochastic environments.
arxiv Mar 7
Neuro-Symbolic Artificial Intelligence: A Task-Directed Survey in the Black-Box Models Era
Discusses neuro-symbolic (NeSy) integration for inferring behavioral schemas, limited by semantic generalizability in complex domains.
arxiv Mar 7
FEAST: Retrieval-Augmented Multi-Hierarchical Food Classification for the FoodEx2 System
Addresses hierarchical text classification challenges in FoodEx2 food descriptions using HTC/XML techniques.
arxiv Mar 7
CMES | Free Full-Text | Digital Twins and Cyber-Physical Systems: A New Frontier in Computer Modeling
Studies use IoT, AI, and big data for seamless digital twin (DT) integration in cyber-physical systems (CPS). AI research relevance: optimizes DT operations via analytics.
web Mar 7
Frontiers | Autonomous cyber-physical security middleware for IoT: anomaly detection and adaptive response in hybrid environments
Combines IDS with Eclipse Ditto DT for real-time detection of attacks like Hping3 and NMAP.
web Mar 7
Artificial intelligence-driven security framework for internet of things-enhanced digital twin networks - ScienceDirect
Highlights privacy, confidentiality, and reliability risks in physical-digital twin data exchanges with IoT.
web Mar 7
SPRINT: Semi-supervised Prototypical Representation for Few-Shot Class-Incremental Tabular Learning
Explores Few-Shot Class-Incremental Learning (FSCIL) for tabular data streams, leveraging abundant unlabeled data unlike vision methods.
arxiv Mar 7
CAMMSR: Category-Guided Attentive Mixture of Experts for Multimodal Sequential Recommendation
Addresses heuristic fusion limitations in multimodal sequential recommendation systems using text and images for personalized content discovery.
arxiv Mar 7
Activation Outliers in Transformer Quantization: Reproduction, Statistical Analysis, and Deployment Tradeoffs
Reproduces and extends analysis of activation outliers causing PTQ accuracy drops in transformers like BERT on QNLI, from 89.66% to 54.33% in W8A8 quantization.
arxiv Mar 7
9 AI Agents for Research and Analysis | MindStudio
Elicit automates literature reviews and data extraction from 125 million academic papers.
web Mar 7
An Efficient Unsupervised Federated Learning Approach for Anomaly Detection in Heterogeneous IoT Networks
Federated learning in IoT faces challenges from data heterogeneity despite privacy benefits of localized training.
arxiv Mar 7
RANGER: Sparsely-Gated Mixture-of-Experts with Adaptive Retrieval Re-ranking for Pathology Report Generation
Critiques transformer-based pathology report generation from WSIs for lacking specialization and introducing noisy retrieval; proposes improvements (implied).
arxiv Mar 7
Dual-Modality Multi-Stage Adversarial Safety Training: Robustifying Multimodal Web Agents Against Cross-Modal Attacks
Analyzes vulnerabilities in multimodal web agents where DOM injections corrupt both screenshot and accessibility tree observations with deceptive narratives, outperforming text-only attacks on MiniWob++.
arxiv Mar 7
Type-Aware Retrieval-Augmented Generation with Dependency Closure for Solver-Executable Industrial Optimization Modeling
Type-aware RAG method translates natural-language requirements into executable optimization code by enforcing types and dependencies. AI relevance: enhances LLM reliability for automated industrial modeling.
arxiv Mar 7
A digital twin-enhanced cybersecurity framework for IoT in healthcare: Applications in industry 4.0 - ScienceDirect
Proposes DT-enhanced framework for real-time monitoring and anomaly detection in healthcare IoT cybersecurity. AI research relevance: integrates DTs with AI for continuous threat detection.
web Mar 7
Digital Twins: The Virtual Powerhouses Reshaping Cybersecurity - Brandefense
Digital twins create data-driven virtual replicas for real-time simulation and optimization of physical systems.
web Mar 7
Advancing Security with Digital Twins: A Comprehensive Survey
Examines LLMs in DT frameworks for hardware security via natural-language understanding and reasoning.
web Mar 7
Digital twin driven smart factories: real time physics based co-simulation using edge a.i. and federated learning | Scientific Reports
Cites sustainable diffusion incentives for GenAI-driven DTs in industrial CPS and federated edge intelligence for secure IoT. AI relevance: incentivizes AI integration in DT security.
web Mar 7
Cyber Security Education by integrating Digital Twins and Generative AI[v1] | Preprints.org
DTs replicate PLCs/IoT for safe security testing, enhanced by GenAI for automated penetration and threat modeling.
web Mar 7
NeuroSkill(tm): Proactive Real-Time Agentic System Capable of Modeling Human State of Mind
NeuroSkill is a real-time, offline edge system using EXG/BCI and text embeddings to model human state of mind via SKILL.md descriptions.
arxiv Mar 7
HoMMI: Learning Whole-Body Mobile Manipulation from Human Demonstrations
HoMMI enables whole-body mobile manipulation learning from robot-free human demos via egocentric sensing, tackling the human-to-robot embodiment gap. AI research relevance: advances imitation learning scalability for mobile robotics.
arxiv Mar 7
Valet: A Standardized Testbed of Traditional Imperfect-Information Card Games
Introduces Valet, a testbed of 21 imperfect-information card games to benchmark AI algorithms' robustness across diverse games.
arxiv Mar 7
ULTRA: Unified Multimodal Control for Autonomous Humanoid Whole-Body Loco-Manipulation
Introduces ULTRA, a unified framework for whole-body loco-manipulation in humanoids that generates behaviors from perception and task specs, overcoming data scarcity and motion tracking limits. Key AI relevance: scales learning for versatile humanoid robotics beyond retargeted data.
arxiv Mar 7
How to Peel with a Knife: Aligning Fine-Grained Manipulation with Human Preference
Presents a learning framework for contact-rich manipulation tasks with implicit, subjective success criteria like food prep, addressing challenges in evaluation and reward engineering.
arxiv Mar 7
FaultXformer: A Transformer-Encoder Based Fault Classification and Location Identification model in PMU-Integrated Active Electrical Distribution System
FaultXformer uses Transformer encoders on PMU time-series data for fault detection in electrical grids with DERs.
arxiv Mar 7
SafeGen-LLM: Enhancing Safety Generalization in Task Planning for Robotic Systems
SafeGen-LLM generates safe robotic task plans that generalize to novel properties across domains, outperforming classical and RL methods.
arxiv Mar 7
Controllable Reasoning Models Are Private Thinkers
Training LLMs to follow privacy instructions in reasoning traces prevents unintended data leakage in agentic systems.
arxiv Mar 7
Benchmarking Multi-Agent AI: Insights & Practical Use | Galileo
Presents a flexible benchmark for multi-agent systems to compare architectural approaches.
web Mar 7
MARS: Reinforcing Multi-Agent Reasoning of LLMs through Self-Play in Strategic Games
Emphasizes developing LLMs for effective cooperation and competition in multi-agent systems toward advanced intelligence.
web Mar 7
Microeconomic Foundations of Multi-Agent Learning
Develops economic theory for multi-agent learning in principal-agent MDPs with strategic externalities.
web Mar 7
The Agent Economy: A Blockchain-Based Foundation for Autonomous AI Agents
Agent Economy uses blockchain to enable AI agents as economic peers with assets and payments.
web Mar 7
[2510.06903] When Machines Meet Each Other: Network Effects and the Strategic Role of History in Multi-Agent AI
Experimental framework tests LLM agents in network-effect games for economic equilibrium predictions.
web Mar 7
7 Unbeatable Tools for Easy Content Curation in 2025
Lists AI-assisted tools like Feedly and Elicit for content discovery, curation, and research.
web Mar 7
Revisiting the Role of Review Articles in the Age of AI-Agents: Integrating AI-Reasoning and AI-Synthesis Reshaping the Future of Scientific Publishing | Bratislava Medical Journal | Springer Nature Link
Highlights AI reducing literature review time from 42 days to 9 hours, enabling dynamic reviews and gap analysis.
web Mar 7
AI-Generated Content in Cross-Domain Applications: Research Trends, Challenges and Propositions
Algorithmic curation systems amplify AI-generated content (AIGC) for viral spread across platforms.
web Mar 7
An Economy of AI Agents
Surveys AI agent developments for long-horizon tasks and highlights open economic questions.
web Mar 7
Socio-Economic Model of AI Agents
LLM agents generate context-based human-like decisions, enabling strong individual simulations in economic experiments.
web Mar 7
From single-agent to multi-agent: a comprehensive review of LLM-based legal agents
Reviews enhancements in legal AI like syllogism prompts, logic benchmarks, retrieve-read frameworks, and emotional interaction.
web Mar 5
Beyond Factual Correctness: Mitigating Preference-Inconsistent Explanations in Explainable Recommendation
Introduces PURE, a select-then-generate framework for preference-consistent explanations in LLM recommenders, addressing inconsistencies missed by standard metrics.
arxiv Mar 5
A Comprehensive Survey on Multi-Agent Cooperative Decision-Making: Scenarios, Approaches, Challenges and Perspectives
Surveys multi-agent system decision-making, prioritizing MARL and LLM-based over traditional rule/game/evolutionary methods.
web Mar 5
On the Expressive Power of Transformers for Maxout Networks and Continuous Piecewise Linear Functions
Demonstrates Transformers approximate maxout networks, inheriting ReLU-like universal approximation with comparable complexity.
arxiv Mar 5
Safe and Robust Domains of Attraction for Discrete-Time Systems: A Set-Based Characterization and Certifiable Neural Network Estimation
Develops a framework for estimating safe, robust domains of attraction in uncertain, constrained nonlinear discrete-time systems.
arxiv Mar 5
AgentIR: Reasoning-Aware Retrival for Deep Research Agents
Introduces Reasoning-Aware Retrieval to leverage explicit natural language reasoning from Deep Research agents, which existing retrievers ignore, enhancing retrieval for AI agents.
arxiv Mar 5
Turning Trust to Transactions: Tracking Affiliate Marketing and FTC Compliance in YouTube's Influencer Economy
Develops tools using AI (implied) to analyze YouTube affiliate marketing disclosures and non-compliance with FTC guidelines.
arxiv Mar 5
Web of Science AI Research Assistant | Clarivate
Web of Science Research Assistant employs agentic AI to simplify literature reviews by scoping topics, refining searches, and curating papers.
web Mar 5
$τ$-Knowledge: Evaluating Conversational Agents over Unstructured Knowledge
Introduces τ-Knowledge benchmark extending τ-Benchmark to evaluate retrieval and tool use in conversational agents over unstructured corpora in long-horizon interactions.
arxiv Mar 5
A Dual-Helix Governance Approach Towards Reliable Agentic AI for WebGIS Development
Proposes a dual-helix governance framework with a 3-track (Knowledge, Behavior, Skills) architecture using knowledge graphs to address LLM limitations like context constraints and forgetting in agentic AI for WebGIS development.
arxiv Mar 5
How AI Automates Knowledge Curation
Magai provides comprehensive AI curation for text, images, and videos, outperforming conventional knowledge systems.
web Mar 5
AI knowledge base: A complete guide for 2026
Tettra uses AI-powered search, tagging, and dashboards to manage knowledge bases, routing unanswered queries to humans and identifying gaps.
web Mar 5
Conversations with AI Part IV: AI-assisted knowledge libraries and curation | Ipsos
Evaluates AI for speeding up document curation while preserving accuracy and human judgment in pilots balancing automation with expert insights.
web Mar 5
The role of AI in content curation
AI algorithms enable web scouring via keyword searches to curate valuable content for audiences.
web Mar 5
Robustness of Agentic AI Systems via Adversarially-Aligned Jacobian Regularization
Introduces Adversarially-Aligned Jacobian Regularization (AAJR) to stabilize minimax training in LLM multi-agent ecosystems by addressing non-linear policy instabilities without conservative global bounds.
arxiv Mar 5
(PDF) Collaborative Multi-Agent Test-Time Reinforcement Learning for Reasoning
Introduces MATTRL, injecting structured textual experience into multi-agent deliberation at inference time.
web Mar 4
Proactive Guiding Strategy for Item-side Fairness in Interactive Recommendation
Proposes proactive fairness in recommenders by guiding user preferences toward long-tail items, avoiding preference misalignment from direct insertion.
arxiv Mar 4
Compact Prompting in Instruction-tuned LLMs for Joint Argumentative Component Detection
Discusses argumentative component detection (ACD) in argument mining as a challenging task, with existing methods simplifying to labeling or pipelines.
arxiv Mar 4
Cybersecurity Digital Twins: Concept, blueprint, and challenges for multi-ownership digital service chains - ScienceDirect
Elaborates a blueprint for Cyber-security Digital Twins to model security postures of interconnected systems.
web Mar 4
Artificial intelligence-driven security framework for internet of things-enhanced digital twin networks - ScienceDirect
Highlights confidentiality, privacy, and reliability risks in physical-digital twin integration with IoT data flows.
web Mar 4
Geometry-based pneumatic actuators for soft robotics
Geometry-based pneumatic actuators (GPAs) enable predictable deformation and complex patterns for safe human-robot interaction.
arxiv Mar 4
ARGUS: Seeing the Influence of Narrative Features on Persuasion in Argumentative Texts
ARGUS framework studies narrative features' impact on persuasion in argumentative discourse using a new annotated ChangeMyView corpus.
arxiv Mar 4
Robust Skills, Brittle Grounding: Diagnosing Restricted Generalization in Vision-Language Action Policies via Multi-Object Picking
Evaluates vision-language action policies on multi-object picking, revealing reliance on object-location correlations rather than robust grounding.
arxiv Mar 4
Artificial Agency Program: Curiosity, compression, and communication in agents
Artificial Agency Program envisions curiosity-driven, resource-bounded AI agents embedded in human-tool systems for enhanced sensing and actuation.
arxiv Mar 4
ArgLLM-App: An Interactive System for Argumentative Reasoning with Large Language Models
Presents ArgLLM-App, a web system using argumentative LLMs for explainable, contestable decisions with human interaction.
arxiv Mar 4
A digital twin-enhanced cybersecurity framework for IoT in healthcare: Applications in industry 4.0 - ScienceDirect
Presents scalable DT-CF architecture with AI-driven analytics for reliable threat mitigation in healthcare via real-time physical-virtual sync.
web Mar 4
Advancing Security with Digital Twins: A Comprehensive Survey
Examines integrating LLMs into digital twins to bolster hardware security via natural-language understanding and reasoning.
web Mar 4
Digital twin driven smart factories: real time physics based co-simulation using edge a.i. and federated learning | Scientific Reports
Cites sustainable diffusion incentives for generative AI-driven digital twins in industrial CPS and federated edge intelligence.
web Mar 4
Odin: Multi-Signal Graph Intelligence for Autonomous Discovery in Knowledge Graphs
Presents Odin, a production graph engine using COMPASS score (PageRank + NPLL) for autonomous pattern discovery in knowledge graphs.
arxiv Mar 4
From Complex Dynamics to DynFormer: Rethinking Transformers for PDEs
Critiques Transformer-based neural operators for uniformly treating spatial points in PDE solving, ignoring scale separation and incurring high costs.
arxiv Mar 4
Beyond Task Completion: Revealing Corrupt Success in LLM Agents through Procedure-Aware Evaluation
Introduces Procedure-Aware Evaluation (PAE) for LLM-based agents, assessing procedures via structured observations across Utility, Efficiency, Interaction Quality, and Procedural Integrity beyond mere task completion.
arxiv Mar 4
Why Adam Can Beat SGD: Second-Moment Normalization Yields Sharper Tails
Proves Adam's superiority over SGD via second-moment normalization under bounded variance using martingale analysis, explaining empirical convergence gaps.
arxiv Mar 4
Multi-Scale Adaptive Neighborhood Awareness Transformer For Graph Fraud Detection
Highlights GNN limitations in graph fraud detection due to homogeneity assumptions and poor global modeling, proposing solutions to these challenges.
arxiv Mar 4
Frontiers | Multi-agent systems powered by large language models: applications in swarm intelligence
Shows 70B LLMs outperform 7B models in resilience to noise or irrelevant data.
web Mar 4
Agentic AI: The age of reasoning—A review - ScienceDirect
Offers a chronological overview of agentic AI milestones, key papers, and breakthroughs.
web Mar 4
LLM-Based Multi-Agent Systems for Mathematical Problem Solving: A Comprehensive Literature Review[v1] | Preprints.org
Details math benchmarks (e.g., MATH500, GSM8K), CoT prompting, RL fine-tuning, and hierarchical multi-agent architecture for reasoning.
web Mar 4
Evaluation and Benchmarking of LLM Agents: A Survey
Argues multi-agent LLM evaluation requires new methods distinct from RL due to lack of predefined rewards.
web Mar 4
Unlocking the Power of Multi-Agent LLM for Reasoning: From Lazy Agents to Deliberation
Analyzes emergence of lazy agents in multi-agent LLM reasoning frameworks, even with strong system performance, calling for online RL solutions.
web Mar 4
[2509.01063] An Economy of AI Agents
AI agents capable of long-horizon complex tasks will deploy across the economy with minimal human oversight.
web Mar 2
A Minimal Agent for Automated Theorem Proving
A minimal agentic baseline enables comparison of AI theorem prover architectures via iterative refinement and library search.
arxiv Mar 2
DARE-bench: Evaluating Modeling and Instruction Fidelity of LLMs in Data Science
DARE-bench introduces a benchmark for LLMs in complex data science tasks, addressing gaps in process-aware evaluation and labeled training data.
arxiv Mar 2
CUDA Agent: Large-Scale Agentic RL for High-Performance CUDA Kernel Generation
LLMs lag in GPU kernel optimization like CUDA generation compared to compilers, with existing refinement methods failing fundamentally.
arxiv Mar 2
Resources for Automated Evaluation of Assistive RAG Systems that Help Readers with News Trustworthiness Assessment
The TREC 2025 DRAGUN track evaluates RAG systems for generating reader-oriented reports on news trustworthiness amid misinformation.
arxiv Mar 2
Efficient Discovery of Approximate Causal Abstractions via Neural Mechanism Sparsification
Structured pruning discovers causal abstractions in neural networks treated as SCMs, verifying interpretable mechanisms without brute-force interventions.
arxiv Mar 2
AI Agents for Economic Research | NBER
Demystifies autonomous LLM-based AI agents for economists with hands-on building instructions for multi-step tasks.
web Mar 2
AgentAI: A comprehensive survey on autonomous agents in distributed AI for industry 4.0 - ScienceDirect
AgentAI transforms distributed AI through autonomous agents operating individually or collaboratively.
web Mar 2
Virtual Agent Economies
Rapid AI agent adoption creates a new economic layer of agent transactions; proposes 'sandbox economy' framework.
web Mar 2
Agentic Inequality
AI agents as autonomous delegates introduce power asymmetries via scalable goal delegation and agent competition.
web Mar 2
Resilient Strategies for Stochastic Systems: How Much Does It Take to Break a Winning Strategy?
Introduces resilient strategies for agents to make robust decisions against disturbances that flip outcomes, like actuator malfunctions, in stochastic settings.
arxiv Mar 2
SafeGen-LLM: Enhancing Safety Generalization in Task Planning for Robotic Systems
SafeGen-LLM addresses safety challenges in robotic task planning by enhancing safety satisfaction and generalizing to novel properties across domains, outperforming classical planners, RL, and base LLMs.
arxiv Mar 2
Leveraging digital twins for advanced threat modeling in cyber-physical systems cybersecurity | International Journal of Information Security | Springer Nature Link
Cites works on intelligent digital twins using spatio-temporal fusion and curriculum learning for IoT attack detection and anomaly in CPS.
web Mar 2
Digital Twins: The Virtual Powerhouses Reshaping Cybersecurity - Brandefense
Digital twin technology creates data-driven virtual replicas for real-time simulation and optimization of physical systems.
web Mar 2
Cyber Security Education by integrating Digital Twins and Generative AI[v1] | Preprints.org
Digital twins replicate PLCs and IoT for proactive security, enhanced by generative AI for tasks like penetration testing.
web Mar 2
The Power of Digital Twins in Cybersecurity – Communications of the ACM
Traditional reactive cybersecurity fails against adaptive AI-powered adversaries in dynamic infrastructures.
web Mar 2
Controllable Reasoning Models Are Private Thinkers
Proposes training AI agents to follow instructions in reasoning traces to prevent unintended leakage of sensitive user data.
arxiv Mar 2
CMES | Free Full-Text | Digital Twins and Cyber-Physical Systems: A New Frontier in Computer Modeling
Explores IoT, AI, and big data for seamless digital twin integration in cyber-physical systems, addressing optimization challenges.
web Mar 2
Evaluating Accuracy of Vine Robot Shape Sensing with Distributed Inertial Measurement Units
Develops shape sensing for soft vine robots using IMU and force sensing to localize sensors and configurations in debris-filled environments.
arxiv Mar 2
Learning Flexible Job Shop Scheduling under Limited Buffers and Material Kitting Constraints
Applies deep reinforcement learning to Flexible Job Shop Scheduling with limited buffers and material kitting for real-world production efficiency.
arxiv Mar 2
Towards a Science of Scaling Agent Systems
Defines multi-agent scaling via agents, coordination, models, and tasks, evaluated on benchmarks like Finance-Agent.
web Feb 22
Reverso: Efficient Time Series Foundation Models for Zero-shot Forecasting
Critiques scaling in time series foundation models for inefficiency despite performance gains, advocating alternatives.
arxiv Feb 22
FAMOSE: A ReAct Approach to Automated Feature Discovery
FAMOSE uses ReAct agents for autonomous feature augmentation and selection in tabular ML, reducing domain expertise needs.
arxiv Feb 22
A Comprehensive Survey on Multi-Agent Cooperative Decision-Making: Scenarios, Approaches, Challenges and Perspectives
Categorizes multi-agent decision-making, prioritizing MARL and LLM-based over traditional methods.
web Feb 22
Pushing the Frontier of Black-Box LVLM Attacks via Fine-Grained Detail Targeting
Addresses gradient instability in black-box attacks on LVLMs due to ViT sensitivity, improving transfer-based methods like M-Attack.
arxiv Feb 22
A.R.I.S.: Automated Recycling Identification System for E-Waste Classification Using Deep Learning
Presents A.R.I.S., a YOLOx-based portable sorter for real-time e-waste material classification to boost recycling efficiency.
arxiv Feb 22
When to Trust the Cheap Check: Weak and Strong Verification for Reasoning
Distinguishes weak (cheap, internal) vs. strong (reliable, external) verification in LLM reasoning loops for trustworthiness.
arxiv Feb 22
[2504.19678] From LLM Reasoning to Autonomous AI Agents: A Comprehensive Review
Comprehensive review bridging LLM reasoning to autonomous AI agents.
web Feb 22
Benchmarking Multi-Agent AI: Insights & Practical Use | Galileo
Benchmark supports diverse agent architectures for multi-agent system comparisons.
web Feb 22
[2601.12538] Agentic Reasoning for Large Language Models
Differentiates in-context vs. post-training reasoning in agentic frameworks across domains like science and robotics.
web Feb 22
[2510.06903] When Machines Meet Each Other: Network Effects and the Strategic Role of History in Multi-Agent AI
Studies GPT-5-based LLM agents in a network-effect game to test convergence to fulfilled expectation equilibria under varied conditions. Contributes to AI agent evaluation in economic games.
web Feb 22
Socio-Economic Model of AI Agents
LLM agents generate context-based decisions like humans, enabling individual simulation in economic experiments beyond rule-based agents.
web Feb 22
Levels of Autonomy for AI Agents Working Paper
Discusses discipline-specific intentionality in agents, e.g., moral in philosophy, utility-maximizing in economics.
web Feb 22
AI Gamestore: Scalable, Open-Ended Evaluation of Machine General Intelligence with Human Games
Proposes dynamic benchmarks mimicking broad human activities to evaluate AI general intelligence, overcoming static benchmark limitations. Relevant for rigorous assessment of human-like AI capabilities.
arxiv Feb 22
The Agent Economy: A Blockchain-Based Foundation for Autonomous AI Agents
Proposes Agent Economy, a blockchain platform granting AI agents legal identity, assets, and payments as human peers. Advances AI by enabling autonomous economic participation.
web Feb 22
Probability-Invariant Random Walk Learning on Gyral Folding-Based Cortical Similarity Networks for Alzheimer's and Lewy Body Dementia Diagnosis
Introduces gyral folding-based brain networks using three-hinge gyri to distinguish Alzheimer's from Lewy body dementia, addressing atlas inconsistencies. Applies AI network analysis to personalized neuroimaging.
arxiv Feb 22
Conditional Flow Matching for Continuous Anomaly Detection in Autonomous Driving on a Manifold-Aware Spectral Space
Deep-Flow uses OT-CFM for unsupervised anomaly detection in AV driving by modeling expert human behavior densities. Advances AI safety validation for scaling rare scenario detection in autonomous vehicles.
arxiv Feb 22
BMC4TimeSec: Verification Of Timed Security Protocols
BMC4TimeSec verifies timed security protocols using SMT-based bounded model checking and multi-agent timed interpreted systems. Contributes to AI formal verification in secure multi-agent environments.
arxiv Feb 22
A Hybrid Federated Learning Based Ensemble Approach for Lung Disease Diagnosis Leveraging Fusion of SWIN Transformer and CNN
Presents a hybrid federated learning ensemble of SWIN Transformer and CNN for lung disease diagnosis using shared medical data.
arxiv Feb 22
An Economy of AI Agents
Surveys AI agents for long-horizon task execution with little oversight, targeting economic deployment.
web Feb 22
A Hybrid Federated Learning Based Ensemble Approach for Lung Disease Diagnosis Leveraging Fusion of SWIN Transformer and CNN
Hybrid federated learning ensemble with SWIN Transformer and CNN diagnoses lung diseases from shared data. Applies transformers and FL to privacy-preserving medical AI.
arxiv Feb 22
The Cascade Equivalence Hypothesis: When Do Speech LLMs Behave Like ASR$\rightarrow$LLM Pipelines?
Speech LLMs implicitly perform ASR, equivalent to Whisper+LLM cascades, confirmed by mechanistic analysis. Key insight into multimodal AI internals for speech processing and representation.
arxiv Feb 22
AI Gamestore: Scalable, Open-Ended Evaluation of Machine General Intelligence with Human Games
Proposes dynamic benchmarks mimicking human general intelligence over static narrow AI tests. Crucial for AGI evaluation amid rapid AI progress.
arxiv Feb 22
FR-GESTURE: An RGBD Dataset For Gesture-based Human-Robot Interaction In First Responder Operations
Introduces gesture dataset for UGV control by first responders in disasters. Supports AI in gesture recognition and human-robot interaction for emergency response.
arxiv Feb 22
Digital Twins: The Virtual Powerhouses Reshaping Cybersecurity - Brandefense
Digital twins create virtual replicas for real-time simulation and optimization. Foundational for AI-driven system modeling and performance analysis.
web Feb 22
Hybrid System Planning using a Mixed-Integer ADMM Heuristic and Hybrid Zonotopes
Hybrid zonotopes with ADMM MIP heuristic enable efficient motion planning for hybrid systems.
arxiv Feb 22
Cybersecurity Digital Twins: Concept, blueprint, and challenges for multi-ownership digital service chains - ScienceDirect
Position paper on Cyber-security Digital Twin for modeling interconnected system security. Advances AI/DT concepts for cybersecurity posture assessment.
web Feb 22
Digital twin driven smart factories: real time physics based co-simulation using edge a.i. and federated learning | Scientific Reports
Cites Gen AI-driven DTs with incentives in CPS and federated edge intelligence. Explores generative AI and FL in industrial DTs.
web Feb 22
A survey on security enhancing Digital Twins: Models, applications and tools - ScienceDirect
Shadow twins in AWS IoT form DTs for physical/cyber components and power sharing.
web Feb 22
Enabling Cyber Security Education through Digital Twins and Generative AI
DTs simulate IT/OT/IoT for real-time cybersecurity monitoring.
web Feb 22
MARS: Margin-Aware Reward-Modeling with Self-Refinement
Proposes difficulty-aware data augmentation for reward models in RLHF/RLAIF to improve alignment without costly human labels.
arxiv Feb 22
Multi-Round Human-AI Collaboration with User-Specified Requirements
Defines counterfactual harm and complementarity principles for conversational AI to reliably aid high-stakes human decisions via user-defined rules.
arxiv Feb 22
Sink-Aware Pruning for Diffusion Language Models
DLMs suffer high inference costs from iterative denoising, and unlike AR LLMs, their attention-sink positions show high variance across generation, invalidating inherited pruning heuristics.
arxiv Feb 22
Evaluation and Benchmarking of LLM Agents: A Survey
Highlights unique evaluation needs for LLM-based multi-agent collaboration vs. traditional RL due to absent predefined rewards.
web Feb 22
Unlocking the Power of Multi-Agent LLM for Reasoning: From Lazy Agents to Deliberation
Analyzes lazy agents in multi-agent LLM frameworks and promotes online RL for balanced contributions.
web Feb 22
From single-agent to multi-agent: a comprehensive review of LLM-based legal agents
Reviews legal AI enhancements like syllogism prompts, logic benchmarks, retrieve-then-read, and emotional interaction.
web Feb 22
LLM-Based Multi-Agent Systems for Mathematical Problem Solving: A Comprehensive Literature Review[v1] | Preprints.org
Details math benchmarks (MATH500 etc.) with hierarchical multi-agent setups using CoT prompting and RL fine-tuning.
web Feb 22
AgentAI: A comprehensive survey on autonomous agents in distributed AI for industry 4.0 - ScienceDirect
Traces AgentAI in gaming from rule-based to adaptive multi-agent systems using RL for emergent behavior.
web Feb 22
CLEF HIPE-2026: Evaluating Accurate and Efficient Person-Place Relation Extraction from Multilingual Historical Texts
HIPE-2026 evaluates person-place relation extraction (at/isAt) from noisy multilingual historical texts.
arxiv Feb 22
Mine and Refine: Optimizing Graded Relevance in E-commerce Search Retrieval
Introduces 'Mine and Refine' contrastive training for semantic embeddings handling graded relevance in e-commerce search with long-tail queries.
arxiv Feb 22
CLEF HIPE-2026: Evaluating Accurate and Efficient Person-Place Relation Extraction from Multilingual Historical Texts
HIPE-2026 is a CLEF lab evaluating person-place relation extraction from noisy multilingual historical texts, focusing on 'at' and 'isAt' relations.
arxiv Feb 22
[2509.01063] An Economy of AI Agents
Predicts deployment of AI agents capable of long-horizon planning and execution with minimal human oversight across the economy. Highlights AI's economic impact and open research questions.
web Feb 22
The Cascade Equivalence Hypothesis: When Do Speech LLMs Behave Like ASR$\rightarrow$LLM Pipelines?
Speech LLMs perform implicit ASR, behaving like Whisper-to-LLM cascades on transcript-solvable tasks, confirmed via testing and mechanistic analysis. Key to AI for revealing multimodal LLM internals.
arxiv Feb 22
AutoNumerics: An Autonomous, PDE-Agnostic Multi-Agent Pipeline for Scientific Computing
AutoNumerics is a multi-agent framework that autonomously designs, implements, debugs, and verifies PDE numerical solvers from natural language. Advances AI in automating scientific computing and solver development.
arxiv Feb 22
FAMOSE: A ReAct Approach to Automated Feature Discovery
FAMOSE is a ReAct-based framework that autonomously generates, refines, and selects optimal features for tabular ML data. Key to AI as it automates feature engineering, reducing reliance on domain expertise.
arxiv Feb 22
AgentAI: A comprehensive survey on autonomous agents in distributed AI for industry 4.0 - ScienceDirect
AgentAI enables autonomous agents to operate individually or collaboratively in distributed AI systems.
web Feb 22
Virtual Agent Economies
Autonomous AI agents create a new 'sandbox economy' layer for transactions beyond human oversight, analyzed via key dimensions. Relevant for studying emergent AI-driven economic systems.
web Feb 22
AI Agents for Economic Research by Anton Korinek :: SSRN
Demystifies autonomous LLM-based AI agents that plan, use tools, and execute multi-step tasks, with instructions for economists to build them.
web Feb 22
AI Agents for Economic Research | NBER
Provides economists with practical instructions for building autonomous LLM-based agents that plan, use tools, and execute multi-step research tasks.
web Feb 22
When to Trust the Cheap Check: Weak and Strong Verification for Reasoning
LLM reasoning uses weak internal verification (e.g., self-consistency) and strong external user feedback, balancing cost and reliability.
arxiv Feb 22
CLEF HIPE-2026: Evaluating Accurate and Efficient Person-Place Relation Extraction from Multilingual Historical Texts
HIPE-2026 is a CLEF lab evaluating person-place relation extraction (at/isAt) from noisy multilingual historical texts.
arxiv Feb 22
Advancing Security with Digital Twins: A Comprehensive Survey
LLMs in digital twins boost hardware security via NLU and reasoning.
web Feb 22
When Vision Overrides Language: Evaluating and Mitigating Counterfactual Failures in VLAs
Vision-Language-Action models fail to follow instructions due to counterfactual failures from dataset biases and vision shortcuts. Key to AI research for exposing grounding challenges in embodied language-to-action systems for robotics.
arxiv Feb 22
When to Trust the Cheap Check: Weak and Strong Verification for Reasoning
LLM reasoning uses weak internal verification (e.g., self-consistency) and costly strong external verification (user feedback). Relevant for improving LLM trustworthiness through scalable verification loops in reasoning AI.
arxiv Feb 22
Stable Asynchrony: Variance-Controlled Off-Policy RL for LLMs
Asynchronous RL for LLMs causes high-variance policy gradients from stale rollouts and heavy-tailed ratios. Advances AI training by addressing efficiency issues in scaling RL for LLM reasoning tasks.
arxiv Feb 22
CMES | Free Full-Text | Digital Twins and Cyber-Physical Systems: A New Frontier in Computer Modeling
Studies use IoT, AI, and big data for digital twin integration in CPS optimization. Highlights AI's role in CPS via digital twins.
web Feb 22
The Power of Digital Twins in Cybersecurity – Communications of the ACM
Digital twins enhance cybersecurity through proactive threat detection and simulation. Relevant for AI-driven security in dynamic threat landscapes.
web Feb 22
Leveraging digital twins for advanced threat modeling in cyber-physical systems cybersecurity | International Journal of Information Security | Springer Nature Link
Cites AI digital twins for IoT attack detection and CPS anomaly detection. Advances AI applications in cybersecurity for IoT/CPS.
web Feb 22
Cyber Security Education by integrating Digital Twins and Generative AI[v1] | Preprints.org
DTs replicate PLCs/IoT for safe security testing, enhanced by Gen AI automation. Integrates generative AI into DTs for cybersecurity operations.
web Feb 22
Graph Neural Model Predictive Control for High-Dimensional Systems
GNN-based dynamics models integrated with MPC enable real-time control of high-dimensional systems like soft robots.
arxiv Feb 22
AgentAI: Autonomous Agents in Distributed AI for Industry 4.0
Comprehensive survey on autonomous AI agents in distributed manufacturing environments.
web Feb 21
An Economy of AI Agents
Examines emergent economic behaviors when AI agents interact in market-like environments.
web Feb 21