AI Research News Feeds for December 28th, 2025

AI RESEARCH PAPERS & ACADEMIC SOURCES

• Towards Understanding Gradient Flow Dynamics of Homogeneous Neural Networks Beyond the Origin
• Optimal Complexity in Byzantine-Robust Distributed Stochastic Optimization with Data Heterogeneity
• Towards Unified Native Spaces in Kernel Methods
• TorchCP: A Python Library for Conformal Prediction
• Hopfield-Fenchel-Young Networks: A Unified Framework for Associative Memory Retrieval
• Identifiability of Causal Graphs under Non-Additive Conditionally Parametric Causal Models
• Fundamental Limits of Membership Inference Attacks on Machine Learning Models
• On the Robustness of Kernel Goodness-of-Fit Tests
• Efficient Online Prediction for High-Dimensional Time Series via Joint Tensor Tucker Decomposition
• Fast Computation of Superquantile-Constrained Optimization Through Implicit Scenario Reduction
• Collaborative likelihood-ratio estimation over graphs
• On the Utility of Equal Batch Sizes for Inference in Stochastic Gradient Descent
• Differentially Private Bootstrap: New Privacy Analysis and Inference Strategies
• Convergence and Sample Complexity of Natural Policy Gradient Primal-Dual Methods for Constrained MDPs
• Differentially Private Multivariate Medians
• VFOSA: Variance-Reduced Fast Operator Splitting Algorithms for Generalized Equations
• Scaling Capability in Token Space: An Analysis of Large Vision Language Model
• Minimax Optimal Two-Sample Testing under Local Differential Privacy
• Jackpot: Approximating Uncertainty Domains with Adversarial Manifolds
• An Asymptotically Optimal Coordinate Descent Algorithm for Learning Bayesian Networks from Gaussian Models
• Convergence Rates for Non-Log-Concave Sampling and Log-Partition Estimation
• A Unified Framework to Enforce, Discover, and Promote Symmetry in Machine Learning
• Infinite-dimensional Mahalanobis Distance with Applications to Kernelized Novelty Detection
• Stable learning using spiking neural networks equipped with affine encoders and decoders
• Efficient Knowledge Deletion from Trained Models Through Layer-wise Partial Machine Unlearning
• General Loss Functions Lead to (Approximate) Interpolation in High Dimensions
• Piecewise deterministic sampling with splitting schemes
• Hierarchical and Stochastic Crystallization Learning: Geometrically Leveraged Nonparametric Regression with Delaunay Triangulation
• Gold-medalist Performance in Solving Olympiad Geometry with AlphaGeometry2
• Decentralized Bilevel Optimization: A Perspective from Transient Iteration Complexity
• Fair Text Classification via Transferable Representations
• Stochastic Interior-Point Methods for Smooth Conic Optimization with Applications
• Revisiting Gradient Normalization and Clipping for Nonconvex SGD under Heavy-Tailed Noise: Necessity, Sufficiency, and Acceleration
• Generalized multi-view model: Adaptive density estimation under low-rank constraints
• (De)-regularized Maximum Mean Discrepancy Gradient Flow
• On Probabilistic Embeddings in Optimal Dimension Reduction
• Physics Informed Kolmogorov-Arnold Neural Networks for Dynamical Analysis via Efficient-KAN and WAV-KAN
• Graph-accelerated Markov Chain Monte Carlo using Approximate Samples
• Online Quantile Regression
• Statistical Inference of Random Graphs With a Surrogate Likelihood Function
• On the Representation of Pairwise Causal Background Knowledge and Its Applications in Causal Inference
• An Augmentation Overlap Theory of Contrastive Learning
• Algorithms for ridge estimation with convergence guarantees
• Talent: A Tabular Analytics and Learning Toolbox
• Inferring Change Points in High-Dimensional Regression via Approximate Message Passing
• Universality of Kernel Random Matrices and Kernel Regression in the Quadratic Regime
• Lexicographic Lipschitz Bandits: New Algorithms and a Lower Bound
• On the Natural Gradient of the Evidence Lower Bound
• Geometry and Stability of Supervised Learning Problems
• Understanding Deep Representation Learning via Layerwise Feature Compression and Discrimination
• Optimal Rates of Kernel Ridge Regression under Source Condition in Large Dimensions
• A Hybrid Weighted Nearest Neighbour Classifier for Semi-Supervised Learning
• Scalable and Adaptive Variational Bayes Methods for Hawkes Processes
• Biological Sequence Kernels with Guaranteed Flexibility
• Unified Discrete Diffusion for Categorical Data
• Reinforcement Learning for Infinite-Dimensional Systems
• Deep Neural Networks are Adaptive to Function Regularity and Data Distribution in Approximation and Estimation
• Generation of Geodesics with Actor-Critic Reinforcement Learning to Predict Midpoints
• Learning-to-Optimize with PAC-Bayesian Guarantees: Theoretical Considerations and Practical Implementation
• Sparse Semiparametric Discriminant Analysis for High-dimensional Zero-inflated Data
• Stochastic Interpolants: A Unifying Framework for Flows and Diffusions
• Efficient Methods for Non-stationary Online Learning
• Decentralized Asynchronous Optimization with DADAO allows Decoupling and Acceleration
• Mixtures of Gaussian Process Experts with SMC^2
• Robust Point Matching with Distance Profiles
• BoFire: Bayesian Optimization Framework Intended for Real Experiments
• Reliever: Relieving the Burden of Costly Model Fits for Changepoint Detection
• Variational Inference for Uncertainty Quantification: an Analysis of Trade-offs
• Are Ensembles Getting Better All the Time?
• An Adaptive Parameter-free and Projection-free Restarting Level Set Method for Constrained Convex Optimization Under the Error Bound Condition
• Operator Learning for Hyperbolic PDEs
• Optimal subsampling for high-dimensional partially linear models via machine learning methods
• Decentralized Sparse Linear Regression via Gradient-Tracking
• Calibrated Inference: Statistical Inference that Accounts for Both Sampling Uncertainty and Distributional Uncertainty
• Relaxed Gaussian Process Interpolation: a Goal-Oriented Approach to Bayesian Optimization