Theoretical analysis of rank selection for low-rank tensor regression with applications to neural network compression and model optimization.
Transformer model for survival prediction from incomplete multimodal medical data using modality-specific representation learning and diffusion.
Test-time adaptation framework combining subtractive and additive approaches for object detection under adverse weather domain shifts.
Decoupled audio transformer architecture inspired by human cognition for efficient self-supervised learning on resource-constrained devices.
Analysis of object discovery in self-supervised Vision Transformers, showing how [CLS] token attention maps contain spurious activations affecting localization.
Continual learning method for object detection under extreme visual sparsity conditions using dual-stage invariant learning.
Higher-order associative memory models combining exponential interactions with sparse pattern storage for improved storage capacity.
Analysis of privacy-accuracy trade-offs in high-dimensional sparse linear regression using differential privacy mechanisms and approximate message passing.
Method for compressing conversational audio context in LLM-based speech recognition systems, studying multimodal context from prior turns for improved ASR.
Mixed-resolution vision transformer with adaptive token allocation for efficient dense feature extraction using coarse-to-fine processing.
Vision network for 2D-3D image-to-point-cloud registration using geometry-aware local alignment and structural synchronization.
Approach for merging multiple LoRA modules while preserving subspace coverage and addressing directional anisotropy to maintain task representation in general-purpose systems.
Semi-structured discrete-time model combining additive predictors with neural networks for mortgage delinquency analysis and default prediction.
Benchmark for evaluating machine unlearning in multimodal models like CLIP, introducing SALMUBench with 60K persona-attribute associations for fine-grained forgetting evaluation.
Method for merging independently fine-tuned LoRA adapters across heterogeneous tasks using null-space compression, addressing classification-regression task combinations.
Tensor-network surrogate for efficient option pricing in portfolio risk management using tensor-train approximation.
Graph-learning algorithm (MED-MAGMA) for fitting Kronecker-sum-structured models with multiplicative noise in genomics applications.
Power-weighted noncentral complex Gaussian distribution for signal processing and communications with non-Gaussian amplitude characteristics.
Generative approach for uncertainty quantification in multimodal supervised learning combining images and text data.
Theoretical analysis of Kantorovich-kernel neural network operators with density results, convergence estimates, and Korovkin theorems.
Diffusion models for reconstructing quantum dot charge stability diagrams to accelerate quantum processor characterization.
Meta-learning framework for human mesh recovery from images using optimization-friendly initializations and uncertainty-aware updates.
UNIFERENCE: discrete-event simulation framework for developing and benchmarking distributed AI inference algorithms across heterogeneous devices and networks.
Conditional Neural Bayes Ratio Estimation (cNBRE) for experimental design optimization in frontier physics, applied to 21-cm radio cosmology.
Low-rank RNN approach using flows to infer neural connectivity structure from population recordings while addressing degeneracy in neural dynamics.
AMALIA: fully open source LLM trained on high-quality European Portuguese data with native evaluation benchmark and improved pt-PT representation.
ALBA: linguistically grounded benchmark for evaluating LLM performance on European Portuguese, addressing underrepresentation in existing benchmarks.
ToothCraft: diffusion model for automated dental crown completion using synthetic training data and contextual generation from incomplete teeth.
Experimental pipeline profiling energy consumption, latency, and quality trade-offs for deploying LLMs on edge devices with hardware constraints.
Study evaluating ML feature compatibility and transferability across malware detection datasets under distribution shifts.
Video benchmark for complex perception reasoning requiring multiple temporally separated visual evidence pieces and compositional logic.
Framework for constructing soft equivariant computer vision models by projecting weights into designed subspaces with theoretical bounds.
Deep symbolic regression using policy gradients with complexity awareness for interpretable data-driven mathematical expression discovery.
Approximate dynamic programming approach for global optimization of expensive black-box functions as alternative to Gaussian process Bayesian optimization.
Projection-free algorithms for online convex optimization with time-varying adversarial constraints and regret bounds.
Theoretical analysis of how iteration order affects convergence and stability in deep neural network training without learning rate schedules.
Methodological commentary on robust predictive modeling under distribution shifts in real-world deployment scenarios.
Task Tokens method adapts behavior foundation models to specific tasks via learnable tokens while preserving zero-shot generalization capabilities.
FastCache accelerates Diffusion Transformer inference through learnable linear approximation and spatial-aware token selection for hidden-state caching.
Defends RAG systems against knowledge poisoning attacks by detecting and mitigating adversarial text injections in external knowledge sources.
Masked training approach for robust arrhythmia detection from digitalized ECG images with temporal asynchrony and missing signal segments.
PepThink-R1 integrates LLMs with chain-of-thought supervised fine-tuning and reinforcement learning for interpretable cyclic peptide design optimization.
Generative model for molecular dynamics trajectories using Markov State Models to accelerate computational protein simulations.
LLMs perform automatic wireless modulation classification via discretized self-supervised candidate retrieval, avoiding distribution shift issues of supervised models.
Control-theoretic framework for LLM activation steering with feedback controllers, connecting empirical steering methods to proportional control theory for safety alignment.
NeST-BO proposes Newton-step targeting Bayesian optimization using Gaussian processes to learn gradient and Hessian information for expensive black-box problems.
Analyzes cryptanalytic model extraction attacks on ReLU-based DNNs with hard-label oracle access and polynomial-time complexity.
Sequence-level TopK (SeqTopK) improves Mixture-of-Experts routing in LLMs by adapting expert assignment per sequence rather than per token without retraining.
Cascading Bandits analyzes decision-making policies for edge inference with multiple models, providing theoretical regret guarantees for Explore-then-Commit and Thompson Sampling approaches.
LiteCache optimizes KVCache memory management for LLM inference using GPU-centric query similarity-driven approach to reduce memory overhead and improve CUDA Graph execution.