Ax Charles Ye, Jasmine Cui, Dylan Hadfield-Menell 3/16/2026

Prompt Injection as Role Confusion

Analysis of prompt injection vulnerabilities traced to role confusion in LLMs, with novel detection probes and mitigation approaches.

Ax Huidong Wu, Haojia Xiang, Jingtong Gao, Xiangyu Zhao, Dengsheng Wu, Jianping Li 3/16/2026

Detecting Miscitation on the Scholarly Web through LLM-Augmented Text-Rich Graph Learning

LLM-augmented graph learning approach for detecting miscitations in scholarly networks using context-aware analysis.

Ax Xinyan Jiang, Wenjing Yu, Di Wang, Lijie Hu 3/16/2026

Global Evolutionary Steering: Refining Activation Steering Control via Cross-Layer Consistency

Training-free method for controlling LLM behavior via cross-layer activation steering with improved consistency.

Ax Ming Lei, Shufan Wu, Christophe Baehr 3/16/2026

HCP-DCNet: A Hierarchical Causal Primitive Dynamic Composition Network for Self-Improving Causal Understanding

Neural network architecture for learning causal reasoning with hierarchical primitives and dynamic composition.

Ax Yiwen Song, Tomas Pfister, Yale Song 3/16/2026

VQQA: An Agentic Approach for Video Evaluation and Quality Improvement

Multi-agent agentic framework for video quality evaluation and iterative improvement across diverse generation tasks.

Ax Jacob Adamczyk, Juan Sebastian Rojas, Rahul V. Kulkarni 3/16/2026

Thermodynamics of Reinforcement Learning Curricula

Framework applying non-equilibrium thermodynamics to formalize curriculum learning in reinforcement learning.

Ax Jacob Adamczyk, Adam Kamoski, Rahul V. Kulkarni 3/16/2026

Maximum Entropy Exploration Without the Rollouts

Reinforcement learning exploration method using maximum entropy without rollouts for efficient state space coverage.

Ax Abhinaba Basu, Pavan Chakraborty 3/16/2026

Budget-Sensitive Discovery Scoring: A Formally Verified Framework for Evaluating AI-Guided Scientific Selection

Formally verified evaluation framework for AI-guided scientific candidate selection with budget constraints, accounting for LLM proposal reliability.

Ax Mohamad Alansari, Naufal Suryanto, Divya Velayudhan, Sajid Javed, Naoufel Werghi, Muzammal Naseer 3/16/2026

SPARROW: Learning Spatial Precision and Temporal Referential Consistency in Pixel-Grounded Video MLLMs

MLLM framework for pixel-level video grounding with improved spatial precision and temporal tracking consistency.

Ax Brian Zhang, Deepti Guntur, Zhiyang Zuo, Abhinav Sharma, Shreyas Chaudhari, Wenlong Zhao, Franck Dernoncourt, Puneet Mathur, Ryan Rossi, Nedim Lipka 3/16/2026

Test-Time Strategies for More Efficient and Accurate Agentic RAG

Test-time optimization strategies for agentic RAG systems to reduce inefficient retrieval and improve accuracy on complex multi-hop questions.

Ax Zheda Mai, Ke Zhang, Fu-En Wang, Zixiao Ken Wang, Albert Y. C. Chen, Lu Xia, Min Sun, Wei-Lun Chao, Cheng-Hao Kuo 3/16/2026

Revisiting Model Stitching In the Foundation Model Era

Research on connecting layers across Vision Foundation Models (CLIP, etc.) to measure representational compatibility via stitching.

Ax Emad Sherif, Iryna Yevseyeva, Vitor Basto-Fernandes, Allan Cook 3/16/2026

Operationalising Cyber Risk Management Using AI: Connecting Cyber Incidents to MITRE ATT&CK Techniques, Security Controls, and Metrics

NLP framework maps cyber incidents to MITRE ATT&CK techniques using automated classification for small enterprise risk management.

Ax Xing Zi, Xinying Zhou, Jinghao Xiao, Catarina Moreira, Mukesh Prasad 3/16/2026

Shattering the Shortcut: A Topology-Regularized Benchmark for Multi-hop Medical Reasoning in LLMs

Topology-regularized benchmark for evaluating LLMs on multi-hop medical reasoning, addressing shortcut learning in knowledge graph navigation.

Ax Mak Ahmad, Andrew Macvean, JJ Geewax, David Karger 3/16/2026

The Perfection Paradox: From Architect to Curator in AI-Assisted API Design

Industrial case study evaluating AI-assisted API design workflow comparing AI-generated vs human-authored API specifications with 16 experts.

Ax Shaolong Li, Lichao Sun, Yongchao Chen 3/16/2026

One-Step Flow Policy: Self-Distillation for Fast Visuomotor Policies

One-Step Flow Policy framework distills generative flow models into single-step robotic policies for low-latency visuomotor control.

Ax Qianggang Ding, Haochen Shi, Luis Castej\'on Lozano, Miguel Conner, Juan Abia, Luis Gallego-Ledesma, Joshua Fellowes, Gerard Conangla Planes, Adam Elwood, Bang Liu 3/16/2026

TRACE: Temporal Rule-Anchored Chain-of-Evidence on Knowledge Graphs for Interpretable Stock Movement Prediction

TRACE framework combines knowledge graphs, temporal rules, and LLM-guided reasoning for interpretable stock movement prediction.

Ax Joong Ho Kim, Nicholas Thai, Souhardya Saha Dip, Dong Lao, Keith G. Mills 3/16/2026

Na\"ive PAINE: Lightweight Text-to-Image Generation Improvement with Prompt Evaluation

Naïve PAINE improves text-to-image generation by evaluating and optimizing prompts for diffusion models to reduce generation cycles.

Ax Siddharth Srikanth, Freddie Liang, Sophie Hsu, Varun Bhatt, Shihan Zhao, Henry Chen, Bryon Tjanaka, Minjune Hwang, Akanksha Saran, Daniel Seita, Aaquib Tabrez, Stefanos Nikolaidis 3/16/2026

Red-Teaming Vision-Language-Action Models via Quality Diversity Prompt Generation for Robust Robot Policies

Q-DIG framework red-teams Vision-Language-Action models via quality diversity prompt generation to improve robotic policy robustness.

Ax Eddie Landesberg 3/16/2026

When LLM Judge Scores Look Good but Best-of-N Decisions Fail

Study showing LLM judge correlation with global metrics can be misleading for best-of-n selection tasks, with implications for deployment.

Ax Himel Ghosh, Nick Elias Werner 3/16/2026

LLM BiasScope: A Real-Time Bias Analysis Platform for Comparative LLM Evaluation

LLM BiasScope is a web platform for real-time bias analysis comparing outputs from multiple LLM providers (Gemini, DeepSeek, Llama, etc).

Ax Alliot Nagle, Jakhongir Saydaliev, Dhia Garbaya, Michael Gastpar, Ashok Vardhan Makkuva, Hyeji Kim 3/16/2026

TERMINATOR: Learning Optimal Exit Points for Early Stopping in Chain-of-Thought Reasoning

Method for learning optimal stopping points in chain-of-thought reasoning to reduce compute overhead from overthinking in large reasoning models.

Ax Alaa Dalaq, Muzammil Behzad 3/16/2026

Spatio-Semantic Expert Routing Architecture with Mixture-of-Experts for Referring Image Segmentation

Mixture-of-experts architecture for referring image segmentation using spatio-semantic expert routing to match diverse reasoning requirements.

Ax Carlos Purves, Pietro Lio' 3/16/2026

CALF: Communication-Aware Learning Framework for Distributed Reinforcement Learning

Framework training distributed RL policies under realistic network conditions including delays and packet loss for edge-cloud deployment.

Ax Vishnu Teja Kunde, Fatemeh Doudi, Mahdi Farahbakhsh, Dileep Kalathil, Krishna Narayanan, Jean-Francois Chamberland 3/16/2026

Reinforcement Learning for Diffusion LLMs with Entropy-Guided Step Selection and Stepwise Advantages

RL training framework for diffusion language models using entropy-guided step selection to optimize non-autoregressive sequence generation.

Ax Zekun Wu, Adriano Koshiyama, Sahan Bulathwela, Maria Perez-Ortiz 3/16/2026

AgentDrift: Unsafe Recommendation Drift Under Tool Corruption Hidden by Ranking Metrics in LLM Agents

Security study revealing unsafe recommendation drift in tool-augmented LLM agents when tools are corrupted, hidden by standard ranking metrics.

Ax Gihoon Kim, Euntai Kim 3/16/2026

Swap-guided Preference Learning for Personalized Reinforcement Learning from Human Feedback

Method for personalized RLHF with user-specific preferences using swap-guided learning to address posterior collapse in variational preference learning.

Ax Zixin Wen, Yifu Cai, Kyle Lee, Sam Estep, Josh Sunshine, Aarti Singh, Yuejie Chi, Wode Ni 3/16/2026

Feynman: Knowledge-Infused Diagramming Agent for Scalable Visual Designs

AI agent pipeline for scalable diagram generation using knowledge-infused multi-modal systems to create high-quality visual designs.

Ax Zesheng Yang, Xi Jiang, Bingzhang Hu, Weili Guan, Runmin Cong, Guo-Jun Qi, Feng Zheng 3/16/2026

Mastering Negation: Boosting Grounding Models via Grouped Opposition-Based Learning

Dataset and training method improving vision-language grounding models' ability to interpret negative semantics and complex expressions.

Ax David C. Flynn 3/16/2026

Literary Narrative as Moral Probe : A Cross-System Framework for Evaluating AI Ethical Reasoning and Refusal Behavior

Framework for evaluating AI ethical reasoning using literary narratives and moral dilemmas to test genuine moral understanding beyond surface responses.

Ax Yu-Yang Qian, Hao-Cong Wu, Yichao Fu, Hao Zhang, Peng Zhao 3/16/2026

When Drafts Evolve: Speculative Decoding Meets Online Learning

Novel approach combining speculative decoding with online learning to improve LLM inference speed by adapting draft models to better approximate target distribution.

Ax Ty Valencia, Burak Barlas, Varun Singhal, Ruchir Bhatia, Wei Yang 3/16/2026

VLM4Rec: Multimodal Semantic Representation for Recommendation with Large Vision-Language Models

Vision-language model framework for multimodal recommendation systems using semantic representation alignment between items and user preferences.

Ax Yushu Li, Wenlong Deng, Jiajin Li, Xiaoxiao Li 3/16/2026

Spend Less, Reason Better: Budget-Aware Value Tree Search for LLM Agents

Research on budget-aware value tree search for LLM agents to optimize token and tool usage during execution.

Ax Jiawei Hao, Zhiwei Hao, Jianyuan Guo, Li Shen, Yong Luo, Han Hu, Dan Zeng 3/16/2026

LightMoE: Reducing Mixture-of-Experts Redundancy through Expert Replacing

Research on reducing memory overhead of Mixture-of-Experts LLMs through expert replacement technique.

Ax Hongyang Chen, Zhongwu Sun, Hongfei Ye, Kunchi Li, Xuemin Lin 3/16/2026

Continual Learning in Large Language Models: Methods, Challenges, and Opportunities

Survey of continual learning methods for LLMs to adapt dynamically while preventing catastrophic forgetting.

Ax Lehui Li, Yuyao Wang, Jisheng Yan, Wei Zhang, Jinliang Deng, Haoliang Sun, Zhongyi Han, Yongshun Gong 3/16/2026

From Text to Forecasts: Bridging Modality Gap with Temporal Evolution Semantic Space

Research paper on fusing textual information with time-series forecasting by bridging modality gaps.

Ax Hanbum Ko, Chanhui Lee, Ye Rin Kim, Rodrigo Hormazabal, Sehui Han, Sungbin Lim, Sungwoong Kim 3/16/2026

RetroReasoner: A Reasoning LLM for Strategic Retrosynthesis Prediction

RetroReasoner uses reasoning LLMs for strategic retrosynthesis prediction in organic chemistry with step-by-step planning.

Ax Shuxin Liu, Ou Wu 3/16/2026

MetaKE: Meta-learning Aligned Knowledge Editing via Bi-level Optimization

MetaKE improves knowledge editing in LLMs via bi-level optimization to align semantic targets with feasible execution regions.

Ax Konstantinos F. Xylogiannopoulos, Petros Xanthopoulos, Panagiotis Karampelas, Georgios A. Bakamitsos 3/16/2026

Experimental evidence of progressive ChatGPT models self-convergence

Empirical study of model collapse in ChatGPT through recursive training on synthetic data, demonstrating progressive self-convergence.

Ax Yue Zhang, Chuanlong Qiu, Xinfa Liao, Yiqun Zhang 3/16/2026

Federated Hierarchical Clustering with Automatic Selection of Optimal Cluster Numbers

Federated hierarchical clustering method automatically determines optimal cluster numbers while preserving privacy across distributed data.

Ax Zhuyu Teng, Pei Chen, Yichen Cai, Ruoqing Lu, Zhaoqu Jiang, Jiayang Li, Weitao You, Lingyun Sun 3/16/2026

Seeing Eye to Eye: Enabling Cognitive Alignment Through Shared First-Person Perspective in Human-AI Collaboration

Eye2Eye framework enables human-AI collaboration through shared first-person perspective to reduce communication and understanding gaps.

Ax Donglin Yu 3/16/2026

Cost-Efficient Multimodal LLM Inference via Cross-Tier GPU Heterogeneity

Optimizes multimodal LLM inference cost by partitioning vision and language components across heterogeneous GPU tiers.

Ax Kaifan Zhang, Lihuo He, Junjie Ke, Yuqi Ji, Lukun Wu, Lizi Wang, Xinbo Gao 3/16/2026

CognitionCapturerPro: Towards High-Fidelity Visual Decoding from EEG/MEG via Multi-modal Information and Asymmetric Alignment

CognitionCapturerPro reconstructs visual stimuli from EEG using multi-modal information and asymmetric alignment.

Ax Chenghan Wu, Zongmin Yu, Boai Sun, Liu Yang 3/16/2026

Graph In-Context Operator Networks for Generalizable Spatiotemporal Prediction

Graph In-Context Operator Networks enable neural networks to infer solution operators from examples for spatiotemporal prediction.

Ax Chenyang Zhu, Hongxiang Li, Xiu Li, Long Chen 3/16/2026

MoKus: Leveraging Cross-Modal Knowledge Transfer for Knowledge-Aware Concept Customization

MoKus performs knowledge-aware concept customization by transferring cross-modal knowledge to bind textual knowledge to visual concepts.

Ax Alexander K Taylor, Junyi Zhang, Ethan Ji, Vigyan Sahai, Haikang Deng, Yuanzhou Chen, Yifan Yuan, Di Wu, Jia-Chen Gu, Kai-Wei Chang, Nanyun Peng, Amit Sahai, Wei Wang 3/16/2026

TaoBench: Do Automated Theorem Prover LLMs Generalize Beyond MathLib?

TaoBench evaluates generalization of automated theorem prover LLMs beyond MathLib using novel definitional frameworks.

Ax Guodong Fan, Shengning Zhou, Genji Yuan, Huiyu Li, Jingchun Zhou, Jinjiang Li 3/16/2026

Empowering Semantic-Sensitive Underwater Image Enhancement with VLM

Vision-Language Models enhance underwater image enhancement by improving semantic cue extraction for downstream vision tasks.

Ax Yerin Kwak, Zachary A. Pardos 3/16/2026

The RIGID Framework: Research-Integrated, Generative AI-Mediated Instructional Design

RIGID Framework integrates research-based knowledge into AI-mediated instructional design workflows for education.

Ax Yichen Zhang, Da Peng, Zonghao Guo, Zijian Zhang, Xuesong Yang, Tong Sun, Shichu Sun, Yidan Zhang, Yanghao Li, Haiyan Zhao, Wang Xu, Qi Shi, Yangang Sun, Chi Chen, Shuo Wang, Yukun Yan, Xu Han, Qiang Ma, Wei Ke, Liang Wang, Zhiyuan Liu, Maosong Sun 3/16/2026

Cheers: Decoupling Patch Details from Semantic Representations Enables Unified Multimodal Comprehension and Generation

arXiv paper presenting Cheers, unified multimodal model decoupling patch details from semantics for image comprehension and generation.

Ax Gyutae Oh, Jungwoo Bae, Jitae Shin 3/16/2026

Residual SODAP: Residual Self-Organizing Domain-Adaptive Prompting with Structural Knowledge Preservation for Continual Learning

arXiv paper on Residual SODAP for continual learning in domain-incremental settings using prompt-based adaptation.

Ax Fuhai Chen, Pengpeng Huang, Junwen Wu, Hehong Zhang, Shiping Wang, Xiaoguang Ma, Xuri Ge 3/16/2026

Hierarchical Dual-Change Collaborative Learning for UAV Scene Change Captioning

arXiv paper proposing UAV Scene Change Captioning task using multimodal learning to describe semantic changes in aerial imagery.