R
Published on

ICLR 2026 — Optimization for ML

Optimization for ML

45 papers (0 oral)

Hyperparameter Trajectory Inference with Conditional Lagrangian Optimal Transport

Multiplayer Nash Preference Optimization

p-lessp\textrm{-less} Sampling: A Robust Hyperparameter-Free Approach for LLM Decoding

AutoEP: LLMs-Driven Automation of Hyperparameter Evolution for Metaheuristic Algorithms

Amortising Inference and Meta-Learning Priors in Neural Networks

DVLA-RL: Dual-Level Vision–Language Alignment with Reinforcement Learning Gating for Few-Shot Learning

FSPO: Few-Shot Optimization of Synthetic Preferences Effectively Personalizes to Real Users

Adaptive Concept Discovery for Interpretable Few-Shot Text Classification

Neural Force Field: Few-shot Learning of Generalized Physical Reasoning

Meta-Adaptive Prompt Distillation for Few-Shot Visual Question Answering

Consistency-Driven Calibration and Matching for Few-Shot Class Incremental Learning

RF-DETR: Neural Architecture Search for Real-Time Detection Transformers

On Optimal Hyperparameters for Differentially Private Deep Transfer Learning

Towards Privacy-Guaranteed Label Unlearning in Vertical Federated Learning: Few-Shot Forgetting Without Disclosure

Low-Rank Few-Shot Node Classification by Node-Level Graph Diffusion

Preserve and Sculpt: Manifold-Aligned Fine-tuning of Vision-Language Models for Few-Shot Learning

Aligner, Diagnose Thyself: A Meta-Learning Paradigm for Fusing Intrinsic Feedback in Preference Alignment

Understanding the Mechanisms of Fast Hyperparameter Transfer

SAFA-SNN: Sparsity-Aware On-Device Few-Shot Class-Incremental Learning with Fast-Adaptive Structure of Spiking Neural Network

Meta-Learning Theory-Informed Inductive Biases using Deep Kernel Gaussian Processes

MAGO: Beyond Fixed Hyperparameters with Multi-Objective Pareto Optimization for Hybrid LLM Reasoning

FSOD-VFM: Few-Shot Object Detection with Vision Foundation Models and Graph Diffusion

Doubly-Robust LLM-as-a-Judge: Externally Valid Estimation with Imperfect Personas

Exploring Cross-Modal Flows for Few-Shot Learning

Instance-wise Adaptive Scheduling via Derivative-Free Meta-Learning

Prompt-MII: Meta-Learning Instruction Induction for LLMs

Point-UQ: An Uncertainty-Quantification Paradigm for Point Cloud Few-Shot Class Incremental Learning

Completed Hyperparameter Transfer across Modules, Width, Depth, Batch and Duration

Dual Distillation for Few-Shot Anomaly Detection

Rating Quality of Diverse Time Series Data by Meta-learning from LLM Judgment

Defending Against Unknown Corrupted Agents: Reinforcement Learning of Adversarially Robust Nash Equilibria

3399. SciTS: Scientific Time Series Understanding and Generation with LLMs

  • Topics: LLMs & Foundation Models, Graphs & Structured Data

Expert Merging in Sparse Mixture of Experts with Nash Bargaining

Foundation Visual Encoders Are Secretly Few-Shot Anomaly Detectors

Identifying Robust Neural Pathways: Few-Shot Adversarial Mask Tuning for Vision-Language Models

SwiftTS: A Swift Selection Framework for Time Series Pre-trained Models via Multi-task Meta-Learning

Nasty Adversarial Training: A Probability Sparsity Perspective for Robustness Enhancement

FastGHA: Generalized Few-Shot 3D Gaussian Head Avatars with Real-Time Animation

Binomial Gradient-Based Meta-Learning for Enhanced Meta-Gradient Estimation

Constraint-guided Hardware-aware NAS through Gradient Modification

Exploiting Low-Dimensional Manifold of Features for Few-Shot Whole Slide Image Classification

Object-Centric World Models from Few-Shot Annotations for Sample-Efficient Reinforcement Learning