Tikhonov regularization
PulseAugur coverage of Tikhonov regularization — every cluster mentioning Tikhonov regularization across labs, papers, and developer communities, ranked by signal.
4 day(s) with sentiment data
-
New research challenges standard neural scaling exponent derivation
A new research paper proposes an alternative to the standard geometric derivation of neural scaling exponents, which typically relies on the intrinsic dimension of a data manifold. The authors demonstrate that for modul…
-
New Hyper-Kernel Ridge Regression Tackles Curse of Dimensionality
Researchers have developed Hyper-Kernel Ridge Regression (HKRR), a novel approach that combines deep neural networks and kernel methods to address the curse of dimensionality in machine learning. This method is designed…
-
New research explores theoretical limits of neural network generalization · 4 papers
Four new research papers delve into the theoretical underpinnings of generalization in neural networks. One paper establishes a necessary and sufficient condition for provable compositional generalization, focusing on s…
-
New law precisely predicts neural network optimization instability
Researchers have identified a precise mathematical law governing the instability of scale-invariant optimization in neural networks. This law reveals a feedback loop between learning-rate schedules and weight decay, med…
-
New FR-PT Framework Enhances Neural Network Post-Training Adaptation
Researchers have introduced Feature-level Reverse Propagation for Post-Training (FR-PT), a novel hierarchical framework designed to enhance transparency and control in neural network adaptation after initial training. T…
-
New research details 'canalization' phenomenon in neural network generalization
Researchers have identified a phenomenon called "canalization" in overparameterized neural networks, which describes how the selection of solutions that fit training data evolves during training. This process, observed …
-
El Niño predictability enhanced by delayed observation models
Researchers have developed models to predict El Niño events using delayed observations of the Niño-3.4 index. By analyzing data up to July 2026, they found that incorporating delayed information significantly improves f…
-
Two arXiv papers analyze statistical inverse problems in AI and ML
Two new research papers submitted to arXiv explore statistical inverse problems within machine learning and artificial intelligence. The first paper focuses on regularization techniques for these problems in non-reflexi…
-
New parameter-free fixed-point algorithms achieve linear convergence
Researchers have developed new parameter-free fixed-point algorithms designed for contractive mappings. These algorithms can automatically identify and leverage hidden contractivity without needing prior knowledge of th…
-
Low-cost reservoir computing model advances sign language recognition
Researchers have developed a novel, low-cost hybrid reservoir computing model for recognizing isolated sign language videos. This model utilizes MediaPipe to extract key body and hand points, which are then processed by…
-
New Gaussian Approximation for Ridge Regression Estimator
Researchers have developed a novel Gaussian approximation for the finite-sample distribution of the ridge regression estimator. This approximation accounts for the estimator's bias-variance trade-off in reducing error a…
-
New framework tackles class imbalance in analytic continual learning
Researchers have developed a new framework called Geometry-Spectral Rectification (GSR) to address challenges in class-incremental learning, particularly for datasets with long-tailed distributions. Existing analytic co…
-
New 'weight-norm criticality' explains AI training instability
Researchers have identified a new critical factor in deep neural network training instability, termed 'weight-norm criticality.' This phenomenon, distinct from the commonly understood 'learning-rate criticality,' arises…
-
New prediction-only distillation technique improves AI model training without labeled data
Researchers have developed a new method called prediction-only distillation (POD) for training AI models when labeled data is unavailable. This technique uses fresh, unlabeled covariates to pseudo-label data, which is t…
-
New framework enhances model selection with domain knowledge
A new paper introduces a theoretical framework for model selection using cross-validation, particularly when domain knowledge is incorporated. The research establishes deviation bounds based on VC dimension for the enti…
-
New AI framework traces training data to symbolic policies
Researchers have developed a new framework called Symbolic Mechanistic Data Attribution (SMDA) to better understand how specific training data influences the high-level behavioral decisions of AI models. Unlike previous…
-
New benchmark uses accelerometry data to predict cardiometabolic risk
Researchers have developed a new benchmark dataset derived from NHANES accelerometry data to evaluate tabular learning methods for predicting cardiometabolic risk. The benchmark, comprising data from 1,381 adults, asses…
-
New deep learning model improves tumor scoring for lung cancer
Researchers have developed a novel distribution-based deep multiple instance learning (MIL) framework to improve the accuracy of tumor proportion scoring (TPS) in non-small-cell lung cancer (NSCLC). This approach addres…
-
Sri Lanka remittance study uses AI to predict economic inflows · 2 sources tracked
A new research paper titled "The Remittance Blueprint: Data-driven Intelligence for Sri Lanka" analyzes 32 years of migration and remittance data from Sri Lanka. The study found that external macroeconomic factors like …
-
Linear models with optimized preprocessing match advanced architectures in time-series forecasting
Researchers propose that optimizing preprocessing, rather than scaling model architectures, can significantly improve time-series forecasting accuracy. Using Ridge regression as a testbed, they found that optimal lookba…