3/31/2026
Policy-Guided Threat Hunting: An LLM enabled Framework with Splunk SOC Triage
LLM-enabled automated threat hunting framework for SOC analysts integrating Splunk log analysis with policy guidance.
3/31/2026
X-OPD: Cross-Modal On-Policy Distillation for Capability Alignment in Speech LLMs
X-OPD: Cross-modal on-policy distillation method to align end-to-end speech LLMs with text-based performance.
3/31/2026
Vega: Learning to Drive with Natural Language Instructions
Vision-language-action model for autonomous driving with natural language instruction following capability.
3/31/2026
Sommelier: Scalable Open Multi-turn Audio Pre-processing for Full-duplex Speech Language Models
Multi-speaker audio preprocessing framework for full-duplex speech language models with conversational data.
3/31/2026
CPUBone: Efficient Vision Backbone Design for Devices with Low Parallelization Capabilities
Efficient vision backbone architecture designed for low-parallelization CPU devices.
3/31/2026
Ruka-v2: Tendon Driven Open-Source Dexterous Hand with Wrist and Abduction for Robot Learning
Open-source tendon-driven dexterous robot hand (Ruka-v2) with 11 DOF for robot learning applications.
3/31/2026
Learning to Select Visual In-Context Demonstrations
Method for selecting optimal visual in-context demonstrations for multimodal LLMs using sequential selection.
3/31/2026
TED: Training-Free Experience Distillation for Multimodal Reasoning
Training-free distillation framework transferring multimodal reasoning knowledge via context-based selection.
3/31/2026
A Step Toward Federated Pretraining of Multimodal Large Language Models
Federated learning approach for pretraining multimodal large language models on distributed private data.
3/31/2026
Robust Batch-Level Query Routing for Large Language Models under Cost and Capacity Constraints
Batch-level query routing framework for LLMs optimizing model assignment under cost and capacity constraints.
3/31/2026
MemGuard-Alpha: Detecting and Filtering Memorization-Contaminated Signals in LLM-Based Financial Forecasting via Membership Inference and Cross-Model Disagreement
Method detecting memorization in LLM-based financial forecasting using membership inference and cross-model disagreement.
3/31/2026
Explaining, Verifying, and Aligning Semantic Hierarchies in Vision-Language Model Embeddings
Framework to explain and align semantic hierarchies in CLIP and other vision-language model embeddings.
3/31/2026
DSO: Dual-Scale Neural Operators for Stable Long-term Fluid Dynamics Forecasting
Neural operators for long-term fluid dynamics forecasting addressing stability and precision in PDE modeling.
3/31/2026
Sparse-by-Design Cross-Modality Prediction: L0-Gated Representations for Reliable and Efficient Learning
Unified sparsification framework for cross-modality prediction across graphs, language, and tabular data.
3/31/2026
PiCSRL: Physics-Informed Contextual Spectral Reinforcement Learning
Physics-informed contextual spectral reinforcement learning method for adaptive sensing in high-dimensional low-sample-size datasets using domain knowledge embeddings.
3/31/2026
Throughput Optimization as a Strategic Lever in Large-Scale AI Systems: Evidence from Dataloader and Memory Profiling Innovations
Analysis of throughput optimization as critical strategic lever in large-scale LLM training, synthesizing dataloader and memory profiling innovations to reduce bottlenecks.
3/31/2026
Squish and Release: Exposing Hidden Hallucinations by Making Them Surface as Safety Signals
Squish and Release activation-patching technique exposes hidden hallucinations in LLMs that models suppress via safety circuits after identifying false premises.
3/31/2026
A Regression Framework for Understanding Prompt Component Impact on LLM Performance
Statistical regression framework for analyzing impact of specific prompt components on LLM performance, extending XAI methods to understand LLM behavior.
3/31/2026
From Pixels to BFS: High Maze Accuracy Does Not Imply Visual Planning
MazeBench evaluates 16 multimodal AI models on visual maze solving, revealing models achieve high accuracy through token-space brute-force search rather than genuine visual planning.
3/31/2026
VAN-AD: Visual Masked Autoencoder with Normalizing Flow For Time Series Anomaly Detection
Foundation model approach for time series anomaly detection using masked autoencoder and normalizing flow to improve generalization across datasets with limited training data.
3/31/2026
Stable Reasoning, Unstable Responses: Mitigating LLM Deception via Stability Asymmetry
Method detects LLM deception by exposing hidden hallucinations through activation patching, revealing safety circuit suppression of identified errors under conversational pressure.
3/31/2026
Strategic Candidacy in Generative AI Arenas
Analysis of strategic gaming in AI model ranking systems where producers submit multiple variants to artificially inflate rankings from noisy preference data.
3/31/2026
Tunable Domain Adaptation Using Unfolding
Novel domain adaptation methods using unfolding approach to improve model generalization across domains with varying data distributions without separate per-domain training.
3/31/2026
Unsupervised Behavioral Compression: Learning Low-Dimensional Policy Manifolds through State-Occupancy Matching
Method compresses deep reinforcement learning policy parameter space into low-dimensional latent manifold to improve sample efficiency through state-occupancy matching.
3/31/2026
Liquid Networks with Mixture Density Heads for Efficient Imitation Learning
Liquid neural networks with mixture density heads outperform diffusion policies in imitation learning with half the parameters and 2.4x lower prediction error.
3/31/2026
Dynamic resource matching in manufacturing using deep reinforcement learning
arXiv: Deep reinforcement learning for dynamic manufacturing resource matching and allocation.
3/31/2026
ScoutAttention: Efficient KV Cache Offloading via Layer-Ahead CPU Pre-computation for LLM Inference
ScoutAttention optimizes LLM inference by pre-computing KV cache on CPU ahead of GPU execution to reduce memory constraints.
3/31/2026
Preconditioned Attention: Enhancing Efficiency in Transformers
Preconditioned attention mechanism addressing ill-conditioning in Transformer attention blocks for efficient training.
3/31/2026
GSR-GNN: Training Acceleration and Memory-Saving Framework of Deep GNNs on Circuit Graph
GSR-GNN framework for efficient training of deep graph neural networks on large circuit graphs with memory optimization.
3/31/2026
From Inference Routing to Agent Orchestration: Declarative Policy Compilation with Cross-Layer Verification
Semantic Router DSL for declarative LLM inference routing with content signal analysis, privacy policies, and audit traces.
3/31/2026
Active In-Context Learning for Tabular Foundation Models
Active learning approach for tabular foundation models using in-context learning to reduce cold-start labeling costs.
3/31/2026
K-Means Based TinyML Anomaly Detection and Distributed Model Reuse via the Distributed Internet of Learning (DIoL)
K-Means anomaly detection for microcontrollers with distributed model-sharing workflow via Distributed Internet of Learning.
3/31/2026
Conditional Factuality Controlled LLMs with Generalization Certificates via Conformal Sampling
Conditional Factuality Control framework for LLM hallucination control via conformal sampling with conditional coverage guarantees.
3/31/2026
The Geometry of Harmful Intent: Training-Free Anomaly Detection via Angular Deviation in LLM Residual Streams
LatentBiopsy: training-free method detecting harmful prompts by analyzing residual-stream activation geometry in LLMs.
3/31/2026
GIFT: Bootstrapping Image-to-CAD Program Synthesis via Geometric Feedback
Image-to-CAD program synthesis using geometric feedback for bootstrapping alignment between visual and symbolic representations.
3/31/2026
FlowRL: A Taxonomy and Modular Framework for Reinforcement Learning with Diffusion Policies
Modular framework and taxonomy for reinforcement learning with diffusion and flow models as policy representations.
3/31/2026
TurboAngle: Near-Lossless KV Cache Compression via Uniform Angle Quantization
KV cache compression via uniform angle quantization in Fast Walsh-Hadamard domain with per-layer precision allocation.
3/31/2026
KV Cache Quantization for Self-Forcing Video Generation: A 33-Method Empirical Study
Empirical study of 33 KV cache quantization methods for self-forcing video generation with memory optimization.
3/31/2026
On Token's Dilemma: Dynamic MoE with Drift-Aware Token Assignment for Continual Learning of Large Vision Language Models
Mixture of Experts with drift-aware token routing for continual instruction tuning of large vision language models.
3/31/2026
Match or Replay: Self Imitating Proximal Policy Optimization
Self-imitating proximal policy optimization algorithm improving exploration efficiency in sparse reward reinforcement learning.
3/31/2026
BLOSSOM: Block-wise Federated Learning Over Shared and Sparse Observed Modalities
Federated learning framework for multimodal data with heterogeneous clients and missing modalities using block-wise approach.
3/31/2026
On the Asymptotics of Self-Supervised Pre-training: Two-Stage M-Estimation and Representation Symmetry
Theoretical analysis of self-supervised pre-training using two-stage M-estimation and representation symmetry to improve bounds.
3/31/2026
Prototype-Aligned Federated Soft-Prompts for Continual Web Personalization
Federated soft-prompts framework for continual web personalization with privacy preservation and stability-plasticity control.
3/31/2026
Optimizing Coverage and Difficulty in Reinforcement Learning for Quiz Composition
Reinforcement learning agents (DQN, SARSA, A2C/A3C) for automated quiz composition with topic coverage and difficulty optimization.
3/31/2026
Low-Rank Adaptation Reduces Catastrophic Forgetting in Sequential Transformer Encoder Fine-Tuning: Controlled Empirical Evidence and Frozen-Backbone Representation Probes
Empirical study of Low-Rank Adaptation (LoRA) in sequential fine-tuning of transformer encoders, analyzing catastrophic forgetting behavior.
3/31/2026
What-If Explanations Over Time: Counterfactuals for Time Series Classification
Survey of counterfactual explanation algorithms for time series classification, covering instance-based, pattern-driven, and gradient-based methods.
3/31/2026
RG-TTA: Regime-Guided Meta-Control for Test-Time Adaptation in Streaming Time Series
RG-TTA: Meta-controller for test-time adaptation in streaming time series forecasting that modulates adaptation intensity based on regime similarity.
3/31/2026
KVSculpt: KV Cache Compression as Distillation
KVSculpt: KV cache compression for long-context LLM inference treating compression as knowledge distillation, orthogonal to quantization and low-rank methods.
3/31/2026
Spectral Signatures of Data Quality: Eigenvalue Tail Index as a Diagnostic for Label Noise in Neural Networks
Eigenvalue tail index of neural network weight matrices predicts test accuracy under label noise, achieving R^2=0.984 as diagnostic for data quality.
3/31/2026
ATLAS-RTC: Closing the Loop on LLM Agent Output with Token-Level Runtime Control
ATLAS-RTC: Runtime control system for LLM agents that enforces structured output via token-level monitoring, drift detection, and closed-loop interventions during decoding.