4/17/2026
Trust and Reliance on AI in Education: AI Literacy and Need for Cognition as Moderators
Studies how students' trust in AI assistants affects appropriate reliance and critical evaluation in educational settings.
4/17/2026
SelfGrader: Stable Jailbreak Detection for Large Language Models using Token-Level Logits
SelfGrader detects jailbreak attacks on LLMs using token-level logits without latency overhead or text generation randomness.
4/17/2026
Evaluating the Formal Reasoning Capabilities of Large Language Models through Chomsky Hierarchy
Evaluates formal reasoning capabilities of LLMs using Chomsky hierarchy benchmarks to systematically assess their understanding of computational complexity.
4/17/2026
ChatSVA: Bridging SVA Generation for Hardware Verification via Task-Specific LLMs
ChatSVA uses task-specific LLMs to generate SystemVerilog Assertions for hardware verification, addressing low functional accuracy and data scarcity.
4/17/2026
Token-Budget-Aware Pool Routing for Cost-Efficient LLM Inference
Token-budget-aware pool routing optimizes vLLM inference by routing requests based on estimated token requirements, reducing resource waste and KV-cache failures in production deployments.
4/17/2026
Grid2Matrix: Revealing Digital Agnosia in Vision-Language Models
Grid2Matrix benchmark revealing vision-language models fail at exhaustive visual detail capture in controlled color-grid tasks.
4/17/2026
A-IO: Adaptive Inference Orchestration for Memory-Bound NPUs
A-IO framework for adaptive LLM inference on memory-bound NPU platforms addressing memory bottlenecks during decoding.
4/17/2026
RoboLab: A High-Fidelity Simulation Benchmark for Analysis of Task Generalist Policies
RoboLab: simulation benchmark for evaluating generalist robotics policies with minimal domain overlap between training and evaluation.
4/17/2026
Think in Sentences: Explicit Sentence Boundaries Enhance Language Model's Capabilities
Research showing explicit sentence boundaries improve LLM capabilities by leveraging natural language structure.
4/17/2026
Optimal Stability of KL Divergence under Gaussian Perturbations
Research on KL divergence stability under Gaussian perturbations beyond Gaussian families for OOD detection.
4/17/2026
THEIA: Learning Complete Kleene Three-Valued Logic in a Pure-Neural Modular Architecture
THEIA: modular neural architecture learning Kleene three-valued logic end-to-end without symbolic solvers.
4/17/2026
Network Effects and Agreement Drift in LLM Debates
Studies how LLMs simulate social behaviors in debate scenarios, analyzing agreement drift and network effects.
4/17/2026
The Devil Is in Gradient Entanglement: Energy-Aware Gradient Coordinator for Robust Generalized Category Discovery
Identifies gradient entanglement problem in generalized category discovery and proposes energy-aware gradient coordinator.
4/17/2026
MixAtlas: Uncertainty-aware Data Mixture Optimization for Multimodal LLM Midtraining
MixAtlas method optimizes data mixture composition for multimodal LLM midtraining with benchmark-targeted recipes.
4/17/2026
Towards Verified and Targeted Explanations through Formal Methods
Formal methods combining XAI and neural network verification for trustworthy explanations with mathematical guarantees in safety-critical domains.
4/17/2026
Graph-Based Fraud Detection with Dual-Path Graph Filtering
Graph-based fraud detection using GNNs with dual-path filtering to handle relation camouflage and heterophilic structures in fraud graphs.
4/17/2026
TOPCELL: Topology Optimization of Standard Cell via LLMs
TOPCELL framework using LLMs to optimize transistor topology in standard cell design, reformulating high-dimensional search as guided generation.
4/17/2026
Optimistic Policy Learning under Pessimistic Adversaries with Regret and Violation Guarantees
Policy learning under adversarial and exogenous environments with regret and safety violation guarantees for multi-agent decision-making systems.
4/17/2026
Awakening Dormant Experts:Counterfactual Routing to Mitigate MoE Hallucinations
Counterfactual routing method for sparse Mixture-of-Experts addressing hallucinations by activating dormant experts on long-tail knowledge.
4/17/2026
Metric-Aware Principal Component Analysis (MAPCA):A Unified Framework for Scale-Invariant Representation Learning
Metric-Aware PCA unified framework for scale-invariant representation learning with spectral bias control via metric matrix parameterization.
4/17/2026
Calibrate-Then-Delegate: Safety Monitoring with Risk and Budget Guarantees via Model Cascades
Calibrate-Then-Delegate framework for LLM safety monitoring using model cascades with risk and budget guarantees via correctable error prediction.
4/17/2026
GUI-Perturbed: Domain Randomization Reveals Systematic Brittleness in GUI Grounding Models
GUI-Perturbed framework revealing systematic brittleness in GUI grounding models through domain randomization of visual scenes and instructions.
4/17/2026
Reinforcement Learning via Value Gradient Flow
Behavior-regularized RL for LLM finetuning using value gradient flow to prevent value over-optimization from out-of-distribution extrapolation.
4/17/2026
Enhancing LLM-based Search Agents via Contribution Weighted Group Relative Policy Optimization
Contribution weighted group relative policy optimization for training LLM-based search agents with improved credit assignment and value estimation.
4/17/2026
Quantum-inspired tensor networks in machine learning models
Tensor network methods from quantum physics integrated into ML models to mitigate exponential complexity via compressed representations.
4/17/2026
Tight Sample Complexity Bounds for Best-Arm Identification Under Bounded Systematic Bias
Sample complexity bounds for best-arm identification in autonomous reasoning with LLM-based surrogate models and systematic evaluation bias.
4/17/2026
Modular Continual Learning via Zero-Leakage Reconstruction Routing and Autonomous Task Discovery
Modular architecture for continual learning using task-specific experts and gatekeeper routing to prevent catastrophic forgetting.
4/17/2026
Scouting By Reward: VLM-TO-IRL-Driven Player Selection For Esports
VLM-to-IRL framework for esports player selection based on learned tactical decision-making patterns.
4/17/2026
On the Expressive Power and Limitations of Multi-Layer SSMs
Theoretical analysis of multi-layer state-space models' expressive power and role of chain-of-thought in compositional tasks.
4/17/2026
CI-CBM: Class-Incremental Concept Bottleneck Model for Interpretable Continual Learning
Class-incremental concept bottleneck model addressing catastrophic forgetting in continual learning with interpretability.
4/17/2026
Predicting Post-Traumatic Epilepsy from Clinical Records using Large Language Model Embeddings
LLM embeddings applied to clinical records for early prediction of post-traumatic epilepsy from traumatic brain injury.
4/17/2026
Generative Augmented Inference
Framework leveraging LLM outputs as auxiliary data for parameter estimation in operations management tasks.
4/17/2026
From Risk to Rescue: An Agentic Survival Analysis Framework for Liquidation Prevention
Autonomous agent framework using survival analysis for DeFi liquidation prevention and risk management.
4/17/2026
ConfLayers: Adaptive Confidence-based Layer Skipping for Self-Speculative Decoding
ConfLayers: Adaptive layer-skipping self-speculative decoding speeds LLM inference via confidence-based pruning.
4/17/2026
ELMoE-3D: Leveraging Intrinsic Elasticity of MoE for Hybrid-Bonding-Enabled Self-Speculative Decoding in On-Premises Serving
ELMoE-3D: MoE-based speculative decoding technique improving LLM serving efficiency on memory-bound hardware.
4/17/2026
Zeroth-Order Optimization at the Edge of Stability
Zeroth-order optimization stability analysis for gradient-free learning and memory-efficient model fine-tuning.
4/17/2026
Mean Flow Policy Optimization
MeanFlow: Few-step flow models replace diffusion for efficient policy representation in online RL.
4/17/2026
Constraint-based Pre-training: From Structured Constraints to Scalable Model Initialization
Constraint-based pre-training paradigm enabling model scaling flexibility across varying deployment sizes.
4/17/2026
Adaptive Test-Time Compute Allocation for Reasoning LLMs via Constrained Policy Optimization
Constrained policy optimization framework allocates test-time compute dynamically across LLM reasoning tasks for efficient inference.
4/17/2026
Does RL Expand the Capability Boundary of LLM Agents? A PASS@(k,T) Analysis
PASS@(k,T) metric analyzes whether RL genuinely expands LLM agent capabilities beyond sampling, focusing on agentic tool use.
4/17/2026
xFODE: An Explainable Fuzzy Additive ODE Framework for System Identification
Proposes xFODE framework combining fuzzy and neural ODEs for interpretable nonlinear system identification with physical meaning.
4/17/2026
Can LLMs Score Medical Diagnoses and Clinical Reasoning as well as Expert Panels?
Evaluates LLM juries against expert clinician panels for scoring medical diagnoses and clinical reasoning on real hospital cases.
4/17/2026
Beyond Importance Sampling: Rejection-Gated Policy Optimization
Proposes Rejection-Gated Policy Optimization replacing importance sampling with learned acceptance gates for improved policy gradient updates.
4/17/2026
Multi-User mmWave Beam and Rate Adaptation via Combinatorial Satisficing Bandits
Applies combinatorial satisficing bandits to mmWave beam and rate adaptation in multi-user wireless systems using ACK/NACK feedback.
4/17/2026
LongAct: Harnessing Intrinsic Activation Patterns for Long-Context Reinforcement Learning
Introduces LongAct framework leveraging intrinsic activation patterns in LLMs for long-context reinforcement learning and improved reasoning capabilities.
4/17/2026
Improving Sparse Autoencoder with Dynamic Attention
Proposes dynamic attention mechanism for sparse autoencoders to improve interpretability of foundation model activations by optimizing sparsity levels per neuron.
4/17/2026
Calibration-Gated LLM Pseudo-Observations for Online Contextual Bandits
Uses LLM predictions as pseudo-observations to improve contextual bandit algorithms during cold-start, combining LinUCB with calibration-gated weight injection.
4/17/2026
DLink: Distilling Layer-wise and Dominant Knowledge from EEG Foundation Models
Knowledge distillation method (DLink) for compressing EEG foundation models while preserving task-relevant semantics for embedded BCI systems.
4/17/2026
When Fairness Metrics Disagree: Evaluating the Reliability of Demographic Fairness Assessment in Machine Learning
Research examining disagreement among fairness metrics in ML evaluation and reliability of demographic fairness assessment.
4/17/2026
No More Guessing: a Verifiable Gradient Inversion Attack in Federated Learning
Research on verifiable gradient inversion attacks in federated learning with certification methods for successful privacy breaches.