惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

A
Arctic Wolf
V
V2EX
P
Proofpoint News Feed
The Hacker News
The Hacker News
GbyAI
GbyAI
G
Google Developers Blog
S
Schneier on Security
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
W
WeLiveSecurity
Security Archives - TechRepublic
Security Archives - TechRepublic
博客园 - Franky
Recent Announcements
Recent Announcements
腾讯CDC
Hacker News - Newest:
Hacker News - Newest: "LLM"
K
Kaspersky official blog
U
Unit 42
Engineering at Meta
Engineering at Meta
J
Java Code Geeks
Google Online Security Blog
Google Online Security Blog
Last Week in AI
Last Week in AI
V
Vulnerabilities – Threatpost
N
News and Events Feed by Topic
O
OpenAI News
量子位
cs.CL updates on arXiv.org
cs.CL updates on arXiv.org
Y
Y Combinator Blog
博客园 - 【当耐特】
Vercel News
Vercel News
Hacker News: Ask HN
Hacker News: Ask HN
T
Tor Project blog
Apple Machine Learning Research
Apple Machine Learning Research
Microsoft Security Blog
Microsoft Security Blog
Exploit-DB.com RSS Feed
Exploit-DB.com RSS Feed
AWS News Blog
AWS News Blog
MongoDB | Blog
MongoDB | Blog
S
Security Affairs
A
About on SuperTechFans
Project Zero
Project Zero
D
Darknet – Hacking Tools, Hacker News & Cyber Security
博客园 - 聂微东
Webroot Blog
Webroot Blog
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
Cloudbric
Cloudbric
T
Tenable Blog
月光博客
月光博客
C
Check Point Blog
宝玉的分享
宝玉的分享
V
Visual Studio Blog
T
The Blog of Author Tim Ferriss
NISL@THU
NISL@THU

cs.LG updates on arXiv.org

Portfolio Optimization Proxies under Label Scarcity and Regime Shifts via Bayesian and Deterministic Students under Semi-Supervised Sandwich Training TOPCELL: Topology Optimization of Standard Cell via LLMs Metric-Aware Principal Component Analysis (MAPCA):A Unified Framework for Scale-Invariant Representation Learning Calibrate-Then-Delegate: Safety Monitoring with Risk and Budget Guarantees via Model Cascades Heat and Matérn Kernels on Matchings When Missing Becomes Structure: Intent-Preserving Policy Completion from Financial KOL Discourse Path-Sampled Integrated Gradients Non-intrusive Learning of Physics-Informed Spatio-temporal Surrogate for Accelerating Design Asynchronous Probability Ensembling for Federated Disaster Detection Scouting By Reward: VLM-TO-IRL-Driven Player Selection For Esports Quantization of Spiking Neural Networks Beyond Accuracy An unsupervised decision-support framework for multivariate biomarker analysis in athlete monitoring Predicting Post-Traumatic Epilepsy from Clinical Records using Large Language Model Embeddings Material-Agnostic Zero-Shot Thermal Inference for Metal Additive Manufacturing via a Parametric PINN Framework Physics-Informed Machine Learning for Pouch Cell Temperature Estimation From Risk to Rescue: An Agentic Survival Analysis Framework for Liquidation Prevention CLion: Efficient Cautious Lion Optimizer with Enhanced Generalization Zeroth-Order Optimization at the Edge of Stability Mean Flow Policy Optimization Gating Enables Curvature: A Geometric Expressivity Gap in Attention A Mechanistic Account of Attention Sinks in GPT-2: One Circuit, Broader Implications for Mitigation Expressivity of Transformers: A Tropical Geometry Perspective Assessing the Performance-Efficiency Trade-off of Foundation Models in Probabilistic Electricity Price Forecasting Wasserstein Formulation of Reinforcement Learning. An Optimal Transport Perspective on Policy Optimization Constraint-based Pre-training: From Structured Constraints to Scalable Model Initialization Learning Ad Hoc Network Dynamics via Graph-Structured World Models Adaptive Test-Time Compute Allocation for Reasoning LLMs via Constrained Policy Optimization Curvature-Aligned Probing for Local Loss-Landscape Stabilization Does RL Expand the Capability Boundary of LLM Agents? A PASS@(k,T) Analysis xFODE+: Explainable Type-2 Fuzzy Additive ODEs for Uncertainty Quantification xFODE: An Explainable Fuzzy Additive ODE Framework for System Identification Multi-User mmWave Beam and Rate Adaptation via Combinatorial Satisficing Bandits Blazing the trails before beating the path: Sample-efficient Monte-Carlo planning DLink: Distilling Layer-wise and Dominant Knowledge from EEG Foundation Models Beyond the Laplacian: Doubly Stochastic Matrices for Graph Neural Networks FedIDM: Achieving Fast and Stable Convergence in Byzantine Federated Learning through Iterative Distribution Matching When Flat Minima Fail: Characterizing INT4 Quantization Collapse After FP32 Convergence Assessing the Potential of Masked Autoencoder Foundation Models in Predicting Downhole Metrics from Surface Drilling Data One-shot learning for the complex dynamical behaviors of weakly nonlinear forced oscillators RL-STPA: Adapting System-Theoretic Hazard Analysis for Safety-Critical Reinforcement Learning Optimal last-iterate convergence in matrix games with bandit feedback using the log-barrier How Embeddings Shape Graph Neural Networks: Classical vs Quantum-Oriented Node Representations Benchmarking Optimizers for MLPs in Tabular Deep Learning Predictions of charge density distributions for nuclei with $Z \geq 8$ ML-based approach to classification and generation of structured light propagation in turbulent media Anomaly Detection in IEC-61850 GOOSE Networks: Evaluating Unsupervised and Temporal Learning for Real-Time Intrusion Detection Polyformer: a generative framework for thermodynamic modeling of polymeric molecules Continual Learning for fMRI-Based Brain Disorder Diagnosis via Functional Connectivity Matrices Generative Replay Combining Bayesian and Frequentist Inference for Laboratory-Specific Performance Guarantees in Copy Number Variation Detection Doubly Outlier-Robust Online Infinite Hidden Markov Model PROXIMA: A Reliability Scoring Framework for Proxy Metrics in Online Controlled Experiments Deployment of AI-Assisted Interventions: Capacity Constraints and Noisy Compliance Timescale Separation Enables Deep Reinforcement Learning Control of Rotating Detonation Engine Mode Transitions Bias in Surface Electromyography Features across a Demographically Diverse Cohort DEEP-GAP: Deep-learning Evaluation of Execution Parallelism in GPU Architectural Performance A Synonymous Variational Perspective on the Rate-Distortion-Perception Tradeoff Differentially Private Conformal Prediction Towards Trustworthy 6G Network Digital Twins: A Framework for Validating Counterfactual What-If Analysis in Edge Computing Resources PUFFIN: Protein Unit Discovery with Functional Supervision Expert-Guided Class-Conditional Goodness-of-Fit Scores for Interpretable Classification with Informative Missingness: An Application to Seismic Monitoring Scalable Model-Based Clustering with Sequential Monte Carlo Nautilus: An Auto-Scheduling Tensor Compiler for Efficient Tiled GPU Kernels Best of both worlds: Stochastic & adversarial best-arm identification Regret Tail Characterization of Optimal Bandit Algorithms with Generic Rewards Unraveling the Mechanism of Drug Binding to SARS-CoV-2 RNA Pseudoknot with Thermodynamics-Driven Machine Learning Learning to Concatenate Quantum Codes Unsupervised feature selection using Bayesian Tucker decomposition MLDAS: Machine Learning Dynamic Algorithm Selection for Software-Defined Networking Security Metric-agnostic Learning-to-Rank via Boosting and Rank Approximation MinShap: A Modified Shapley Value Approach for Feature Selection Optimal algorithmic complexity of inference in quantum kernel methods Low-Cost System for Automatic Recognition of Driving Pattern in Assessing Interurban Mobility using Geo-Information A Nonlinear Separation Principle: Applications to Neural Networks, Control and Learning Cloning is as Hard as Learning for Stabilizer States Structural interpretability in SVMs with truncated orthogonal polynomial kernels Kernel Neural Operators (KNOs) for Scalable, Memory-efficient, Geometrically-flexible Operator Learning Logo-LLM: Local and Global Modeling with Large Language Models for Time Series Forecasting Generalization in LLM Problem Solving: The Case of the Shortest Path Diagnosing LLM Judge Reliability: Conformal Prediction Sets and Transitivity Violations Prism: Symbolic Superoptimization of Tensor Programs SegWithU: Uncertainty as Perturbation Energy for Single-Forward-Pass Risk-Aware Medical Image Segmentation Stability and Generalization in Looped Transformers Context Over Content: Exposing Evaluation Faking in Automated Judges AdaSplash-2: Faster Differentiable Sparse Attention MambaSL: Exploring Single-Layer Mamba for Time Series Classification An Analysis of Regularization and Fokker-Planck Residuals in Diffusion Models for Image Generation Class Unlearning via Depth-Aware Removal of Forget-Specific Directions LLMs Gaming Verifiers: RLVR can Lead to Reward Hacking Structure as Computation: Developmental Generation of Minimal Neural Circuits Amortized Optimal Transport from Sliced Potentials IUQ: Interrogative Uncertainty Quantification for Long-Form Large Language Model Generation Beyond Independent Frames: Latent Attention Masked Autoencoders for Multi-View Echocardiography Atropos: Improving Cost-Benefit Trade-off of LLM-based Agents under Self-Consistency with Early Termination and Model Hotswap No More Guessing: a Verifiable Gradient Inversion Attack in Federated Learning When Fairness Metrics Disagree: Evaluating the Reliability of Demographic Fairness Assessment in Machine Learning Route to Rome Attack: Directing LLM Routers to Expensive Models via Adversarial Suffix Optimization What Is the Minimum Architecture for Prolepsis? Early Irrevocable Commitment Across Tasks in Small Transformers Towards Faster Language Model Inference Using Mixture-of-Experts Flow Matching Calibration-Gated LLM Pseudo-Observations for Online Contextual Bandits STEP-Parts: Geometric Partitioning of Boundary Representations for Large-Scale CAD Processing
Eigen-Spike Emergence and Quadratic Equivalents for Conjugate Kernels on Nonlinearly Separable Data
[Submitted on 28 May 2026 (v1), last revised 16 Jun 2026 (this v · 2026-05-29 · via cs.LG updates on arXiv.org

View PDF HTML (experimental)

Abstract:Recent work in random matrix theory (RMT) has developed the notion of deterministic equivalents: typically linear surrogate models that approximate the spectral behavior of large nonlinear random matrices, such as nonlinear feature maps in neural networks (NNs). Such equivalents make theoretical predictions tractable by reducing a complex model to a simpler one with properties that fall under the umbrella of classical RMT tools. However, this leaves open the question of whether this idealized linear equivalence remains meaningful for classification of high-dimensional nonlinearly separable data. Motivated by this, we consider the conjugate kernel (CK), which is the nonlinear feature map of a one-layer feedforward NN, under a canonical nonlinearly separable dataset for the XOR problem; and we use the study of informative outlier eigenvalues in the CK and whether their corresponding eigenvectors asymptotically align with XOR labels as a proxy for nonlinear learnability. We develop a robust quadratic equivalent of the CK matrix that enables a precise analysis of emergent informative spikes, as one modifies various knobs common in ML practice: sample complexity, signal-to-noise ratio (SNR), nonlinear activation choice, and pretrained features. We identify regimes in which these knobs move the CK beyond the linear equivalent and produce BBP-type transitions to label-aligned outlier eigenspaces. Our analysis helps bring deterministic-equivalence tools from RMT to bear on problems of practical relevance in ML.

Submission history

From: Zhichao Wang [view email]
[v1] Thu, 28 May 2026 09:32:19 UTC (6,030 KB)
[v2] Tue, 16 Jun 2026 14:18:22 UTC (6,328 KB)