Skip to content

Latest

arXiv cs.LG Papers

The Granularity Paradox: How Temporal Disaggregation Inflates In-Sample Fit and Compounds Out-of-Sample Error

arXiv:2607.05450v1 Announce Type: new Abstract: This paper explores the "Granularity Paradox" in time-series forecasting, wherein finer temporal disaggregation (e.g., Monthly to Weekly/Daily) improves in-sample diagnostics and…

arXiv cs.LG Papers

Geometry-Aware Infrastructure-Anchored Denoiser for UWB Sensing and Work-Zone Reconstruction

arXiv:2607.05449v1 Announce Type: new Abstract: Accurate work-zone geometry perception is critical for intelligent transportation systems, and ultra-wideband sensing offers a low-cost approach for infrastructure-aided reconstruction. However,…

arXiv cs.LG Papers

Variational Learning of Disentangled Representations

arXiv:2506.17182v3 Announce Type: replace Abstract: Disentangled representations separate factors that are shared across conditions from those that are condition-specific. Such separation is needed for generalization to…

arXiv cs.LG Papers

Breaking Structural Isolation: Scalable Graph Clustering via Community-Aware Sampling and Structural Entropy

arXiv:2607.05469v1 Announce Type: new Abstract: Unsupervised graph clustering is a fundamental technique for uncovering underlying semantic patterns in large-scale networks. Although Graph Contrastive Learning has demonstrated…

arXiv cs.LG Papers

SmartMixed: A Two-Phase Training Strategy for Adaptive Activation Function Learning in Neural Networks

arXiv:2510.22450v4 Announce Type: replace Abstract: The choice of activation function plays a critical role in neural networks, yet most architectures still rely on fixed, uniform activation…

arXiv cs.LG Papers

Universality of Benign Overfitting in Binary Linear Classification

arXiv:2501.10538v3 Announce Type: replace Abstract: The practical success of deep learning has led to the discovery of several surprising phenomena. One of these phenomena, that has…

arXiv cs.LG Papers

Parameter-Free Encoders Remain Viable for RDB Foundation Models

arXiv:2607.05476v1 Announce Type: new Abstract: Given a relational database (RDB) storing heterogeneous tabular information, how can we predict missing (or future) values in some target column…

arXiv cs.LG Papers

Multimodal Analytics of Cybersecurity Crisis Preparation Exercises: What Predicts Success?

arXiv:2603.28553v2 Announce Type: replace-cross Abstract: Instructional alignment, the match between intended cognition and enacted activity, is central to effective instruction but hard to operationalize at scale.…

arXiv cs.LG Papers

Transformers converge to invariant algorithmic cores

arXiv:2602.22600v2 Announce Type: replace Abstract: Training selects for behavior, not circuitry: many weight configurations can implement the same function. Studying any single trained neural network thus…

arXiv cs.LG Papers

PIEFS: Physics-Informed Eigenfunction Features with Learnable Scaling

arXiv:2607.03692v2 Announce Type: replace Abstract: Spectral methods are widely used to construct representations from the geometry of data, but they often rely on a fixed kernel,…