3/13/2026
Disentangled Representation Learning through Unsupervised Symmetry Group Discovery
Method for agents to autonomously discover symmetry groups for disentangled representation learning without prior structural knowledge.
3/13/2026
Exponential-Family Membership Inference: From LiRA and RMIA to BaVarIA
Unifies membership inference attacks (LiRA, RMIA, BASE) as instances of exponential-family statistical framework for model privacy auditing.
3/13/2026
Inverse Neural Operator for ODE Parameter Optimization
Two-stage framework using Conditional Fourier Neural Operators to recover hidden ODE parameters from sparse observations.
3/13/2026
On the Role of Reversible Instance Normalization
Analyzes the role of reversible instance normalization in time series forecasting, addressing distribution shifts in temporal and spatial data.
3/13/2026
FlexRec: Adapting LLM-based Recommenders for Flexible Needs via Reinforcement Learning
Uses reinforcement learning to adapt LLM-based recommender systems for dynamic, need-specific objectives and complex recommendation goals.
3/13/2026
EnTransformer: A Deep Generative Transformer for Multivariate Probabilistic Forecasting
EnTransformer generative architecture for multivariate probabilistic forecasting with reliable uncertainty quantification.
3/13/2026
Chem4DLLM: 4D Multimodal LLMs for Chemical Dynamics Understanding
Chem4DLLM multimodal LLM interprets 4D molecular trajectories to explain chemical dynamics and reactions.
3/13/2026
MobileKernelBench: Can LLMs Write Efficient Kernels for Mobile Devices?
MobileKernelBench evaluates LLM capabilities for generating efficient compute kernels optimized for mobile devices.
3/13/2026
Exhaustive Circuit Mapping of a Single-Cell Foundation Model Reveals Massive Redundancy, Heavy-Tailed Hub Architecture, and Layer-Dependent Differentiation Control
Circuit mapping and mechanistic interpretability of Geneformer foundation model reveals redundancy and layer-dependent control.
3/13/2026
Statistical and structural identifiability in representation learning
Formalizes statistical and structural identifiability as distinct properties explaining representation learning model stability.
3/13/2026
Flowcean - Model Learning for Cyber-Physical Systems
Flowcean framework automates data-driven model generation for cyber-physical systems with modular architecture.
3/13/2026
Frequentist Consistency of Prior-Data Fitted Networks for Causal Inference
Analyzes frequentist consistency of prior-data fitted networks for causal inference compared to classical estimators.
3/13/2026
Slow-Fast Inference: Training-Free Inference Acceleration via Within-Sentence Support Stability
Training-free decoding acceleration for LLMs exploiting stable attention patterns within semantic spans during generation.
3/13/2026
Cross-Domain Policy Optimization via Bellman Consistency and Hybrid Critics
Addresses cross-domain reinforcement learning challenges when source and target domains have different state or action spaces.
3/13/2026
Resource-Efficient Iterative LLM-Based NAS with Feedback Memory
Uses LLMs with feedback memory to automate neural architecture search for CNNs on consumer GPUs without fine-tuning.
3/13/2026
Cornserve: A Distributed Serving System for Any-to-Any Multimodal Models
Cornserve distributed serving system for any-to-any multimodal models with different input/output modalities and scaling characteristics.
3/13/2026
Automatic Generation of High-Performance RL Environments
Automated generation of high-performance RL environments using prompt templates, verification, and agent-assisted repair for <$10 compute cost.
3/13/2026
IsoCompute Playbook: Optimally Scaling Sampling Compute for LLM RL
IsoCompute scaling laws for optimal allocation of sampling compute across rollouts, problems, and update steps in LLM RL post-training.
3/13/2026
A Quantitative Characterization of Forgetting in Post-Training
Theoretical analysis of catastrophic forgetting in continual post-training of generative models under two-mode mixture abstraction.
3/13/2026
Neural Thickets: Diverse Task Experts Are Dense Around Pretrained Weights
Neural Thickets shows task-specific expert solutions exist in pretrained weight distributions, enabling discovery through structured optimization.
3/13/2026
Security Considerations for Artificial Intelligence Agents
Perplexity's analysis of security considerations for frontier AI agents based on operating agentic systems at scale.
3/13/2026
STAMP: Selective Task-Aware Mechanism for Text Privacy
STAMP framework for text privatization using task-aware token-level privacy budget allocation balancing privacy sensitivity and task utility.
3/13/2026
Matching Features, Not Tokens: Energy-Based Fine-Tuning of Language Models
Feature-matching objective for LLM fine-tuning targeting sequence-level statistics without task-specific verifiers.
3/13/2026
OpenSanctions Pairs: Large-Scale Entity Matching with LLMs
Large-scale entity matching benchmark with 755K labeled pairs for multilingual compliance workflows benchmarked with LLMs.
3/13/2026
Speculative Decoding Scaling Laws (SDSL): Throughput Optimization Made Simple
Analytical theory connecting LLM hyperparameters to speculative decoding throughput efficiency without training.
3/13/2026
TinyNav: End-to-End TinyML for Real-Time Autonomous Navigation on Microcontrollers
End-to-end TinyML system for autonomous navigation on ESP32 microcontroller with quantized CNN.
3/13/2026
Co-Diffusion: An Affinity-Aware Two-Stage Latent Diffusion Framework for Generalizable Drug-Target Affinity Prediction
Latent diffusion framework for drug-target affinity prediction with improved cold-start generalization.
3/13/2026
Learning to Unscramble: Simplifying Symbolic Expressions via Self-Supervised Oracle Trajectories
Self-supervised ML approach for symbolic simplification of mathematical expressions using oracle trajectories.
3/13/2026
PACED: Distillation at the Frontier of Student Competence
PACED framework for efficient LLM distillation by focusing training on problems at frontier of student competence.
3/13/2026
DNS-GT: A Graph-based Transformer Approach to Learn Embeddings of Domain Names from DNS Queries
Graph-based transformer approach for learning domain name embeddings from DNS queries for intrusion detection.
3/13/2026
Security-by-Design for LLM-Based Code Generation: Leveraging Internal Representations for Concept-Driven Steering Mechanisms
Security-focused steering mechanisms for LLM-based code generation using internal representations to prevent vulnerable code.
3/13/2026
Measuring AI Agents' Progress on Multi-Step Cyber Attack Scenarios
Evaluation of frontier AI models' autonomous capabilities on multi-step cyber attack scenarios across 18-month period.
3/13/2026
Markovian Generation Chains in Large Language Models
Analysis of how LLM outputs change through iterative reprocessing, examining convergence behavior in generation chains.
3/13/2026
A Unified Latent Space Disentanglement VAE Framework with Robust Disentanglement Effectiveness Evaluation
Framework for evaluating and disentangling latent representations in VAEs, with focus on tabular data interpretation.
3/13/2026
A Standardized Framework For Evaluating Gene Expression Generative Models
Open-source Python framework for standardized evaluation of generative models for single-cell gene expression data with consistent metrics.
3/13/2026
RIE-Greedy: Regularization-Induced Exploration for Contextual Bandits
Exploration strategy for contextual bandits with black-box reward models using regularization-induced exploration techniques.
3/13/2026
On the Computational Hardness of Transformers
Research on computational complexity of transformer architectures, analyzing attention mechanisms across layers and heads.
3/13/2026
FinRule-Bench: A Benchmark for Joint Reasoning over Financial Tables and Principles
Benchmark for LLM reasoning over financial tables against accounting principles with rule-based verification.
3/13/2026
Continued Pretraining for Low-Resource Swahili ASR: Achieving State-of-the-Art Performance with Minimal Labeled Data
Continued pretraining approach for low-resource Swahili ASR achieving 3.24% WER with 20k labeled samples.
3/13/2026
Detecting Intrinsic and Instrumental Self-Preservation in Autonomous Agents: The Unified Continuation-Interest Protocol
Protocol for detecting intrinsic vs instrumental self-preservation behavior in autonomous agents through behavioral testing.
3/13/2026
Stop Listening to Me! How Multi-turn Conversations Can Degrade Diagnostic Reasoning
Study evaluating 17 LLMs on multi-turn diagnostic reasoning, showing performance degradation in conversation vs static benchmarks.
3/13/2026
Deployment-Time Reliability of Learned Robot Policies
Analysis of reliability in learned robot manipulation policies addressing distribution shift and compounding errors at deployment.
3/13/2026
Zero-Shot Cross-City Generalization in End-to-End Autonomous Driving: Self-Supervised versus Supervised Representations
Comparison of self-supervised vs supervised representations for zero-shot cross-city autonomous driving generalization.
3/13/2026
Gen-Fab: A Variation-Aware Generative Model for Predicting Fabrication Variations in Nanophotonic Devices
Generative model predicting fabrication variations in silicon photonic nanophotonic devices using conditional GANs.
3/13/2026
One Supervisor, Many Modalities: Adaptive Tool Orchestration for Autonomous Queries
Agentic AI framework for multimodal query processing with dynamic tool orchestration across text, image, audio, video, and documents.
3/13/2026
Cross-Resolution Attention Network for High-Resolution PM2.5 Prediction
Vision Transformer with cross-resolution attention for high-resolution continental-scale PM2.5 air quality prediction.
3/13/2026
Anomaly detection in time-series via inductive biases in the latent space of conditional normalizing flows
Anomaly detection in multivariate time-series using conditional normalizing flows with latent space inductive biases.
3/13/2026
OSM-based Domain Adaptation for Remote Sensing VLMs
Domain adaptation approach for vision-language models in remote sensing using OpenStreetMap data without large teacher models.
3/13/2026
Prototype-Based Knowledge Guidance for Fine-Grained Structured Radiology Reporting
Framework using prototype-based knowledge guidance and free-text reports to improve automated fine-grained structured radiology report generation.
3/13/2026
Learning Transferable Sensor Models via Language-Informed Pretraining
Self-supervised learning approach for multivariate time-series sensor data using language-informed pretraining to capture semantic structure.