New findings show fixed-kernel discriminators are weaker than feature-learning ones.
problem Comparing performance of fixed-kernel and feature-learning discriminators.
method Using function classes F2 and F1, constructing pairs of distributions, and linking IPMs with sliced Wasserstein distances. result Fixed-kernel IPM and SD cannot discriminate certain distributions that feature-learning IPM and SD can.
Gradient descent learns useful features even in the NTK regime.
problem The ability of neural networks to learn useful features.
method Local convergence analysis of gradient descent with regularization.
result Gradient descent can capture ground-truth directions for feature learning even after the loss threshold is reached.
Study shows optimal model performance at critical level of feature learning.
problem Catastrophic forgetting in neural networks, especially in non-stationary environments.
method Systematic study on model scale and feature learning, using dynamical mean field theory.
result Optimal performance achieved at a critical level of feature learning, dependent on task non-stationarity and model scale.
Feature Learning aims to extract relevant information contained in data sets in an automated fashion. It is driving force behind the current deep learning trend, a set of methods that have had widespread empirical success. What is lacking is a theoretical understanding of different feature learning schemes. This work p…
Neural networks outperform kernels by learning features better.
problem Current theories of feature learning do not adequately assess feature quality.
method Introduced feature quality metric and examined existing theories empirically.
result Current theories of feature learning do not provide a sufficient foundation for neural network generalization.
Theory captures feature learning effects in finite CNNs.
problem Feature learning in finite deep neural networks.
method Derive a self-consistent Gaussian Process theory.
result Good agreement with experiments and sharp transition between regimes.
Multi-task sparse feature learning aims to improve the generalization performance by exploiting the shared features among tasks. It has been successfully applied to many applications including computer vision and biomedical informatics. Most of the existing multi-task sparse feature learning algorithms are formulated a…
Theory explains how deep nets learn features from data.
problem Understanding how deep neural networks learn features from data.
method Developed a noise-nonlinearity phase diagram and a mechanical theory.
result Links feature learning across layers to generalization.
Model shows feature learning can improve neural scaling laws for hard tasks.
problem Understanding and improving neural network scaling laws for various task difficulties.
method Developed a solvable model of neural scaling laws, identified three scaling regimes, and demonstrated feature learning's impact on scaling exponents.
result Feature learning can improve scaling with training time and compute for hard tasks, nearly doubling the exponent.
Exact solutions reveal how unbalanced initializations promote rapid feature learning in neural networks.
problem Understanding how neural networks efficiently extract features from data.
method Deriving exact solutions to a minimal model of neural networks transitioning between lazy and rich learning regimes.
result Unbalanced layer-specific initialization variances and learning rates determine the degree of feature learning.
Deep neural networks are a powerful tool for feature learning and extraction given their ability to model high-level abstractions in highly complex data. One area worth exploring in feature learning and extraction using deep neural networks is efficient neural connectivity formation for faster feature learning and extr…
Paper identifies how neural networks learn features.
problem Understanding how neural networks automatically select features.
method Deep Neural Feature Ansatz, implementing average gradient outer product.
result Recursive Feature Machines achieve state-of-the-art performance.
This paper introduces a new feature learning technique based on error representation.
problem Learning high-level features for classification from diverse and imbalanced data.
method Inverse feature learning using error representation approach.
result Significantly better performance compared to state-of-the-art techniques.
This paper proposes a novel generic one-class feature learning method based on intra-class splitting. In one-class classification, feature learning is challenging, because only samples of one class are available during training. Hence, state-of-the-art methods require reference multi-class datasets to pretrain feature …
Wide neural networks converge linearly to zero loss with feature learning.
problem Optimizing wide neural networks with feature learning guarantees.
method Gradient flow analysis for wide shallow and multi-layer NNs.
result Training loss converges linearly to zero for wide NNs under GF, demonstrating feature learning and better generalization.
Bayesian neural networks explore rare fluctuations for better feature learning.
problem Understanding rare but dominant fluctuations in Bayesian neural networks.
method Large-deviation theory and joint optimization over predictors and internal kernels.
result Posterior rate function optimization reveals data-dependent kernel selection.
New insights into how neural networks learn features, especially when they are very wide.
problem Understanding how gradient flow in wide neural networks selects solutions, especially in the feature-learning regime.
method Axiomatizing the canonical regularizer as a function-space energy and lift, and deriving geodesic ridge for the feature-learning regime.
result Gradient flow in feature-learning networks biases towards ridge regularization, distorting the inductive bias and damaging pretrained networks.
Study on fluctuations in neural network kernels and predictions, focusing on finite width effects.
problem Characterizing fluctuations in finite width neural networks.
method Dynamical mean field theory analysis of wide but finite feature learning neural networks.
result Fluctuations in kernels and predictions are dynamically coupled, leading to reduced variance in feature learning regimes.
Toy model study shows resampling/reweighting can improve feature learning in imbalanced classification.
problem Improving feature learning in imbalanced classification problems.
method High-dimensional toy model with replica method, class-wise resampling/reweighting, and simplified model.
result No resampling/reweighting can sometimes give best feature learning performance.
This work bridges two views of feature learning in neural networks.
problem The relationship between kernel scale changes and data-adaptive feature learning in neural networks remains unresolved.
method Using statistical mechanics, the work derives analytical expressions for network output statistics across scaling regimes.
result Kernel adaptation can be reduced to an effective kernel rescaling, but multi-scale adaptive approach provides richer insights.
Adversarial robustness in multi-index models is as easy as standard learning.
problem Adversarial robustness in high-dimensional multi-index models.
method Proves that hidden directions of multi-index models offer a Bayes optimal low-dimensional projection for robustness against ℓ2-bounded adversarial perturbations. result Adversarially robust learning is as easy as standard learning, requiring no additional samples.
We propose Deep Asymmetric Multitask Feature Learning (Deep-AMTFL) which can learn deep representations shared across multiple tasks while effectively preventing negative transfer that may happen in the feature sharing process. Specifically, we introduce an asymmetric autoencoder term that allows reliable predictors fo…
Diffusion models learn balanced data representations, unlike classification models.
problem Understanding feature learning in diffusion models.
method Proposed a feature learning framework to analyze diffusion models' training dynamics.
result Diffusion models encourage learning balanced and comprehensive representations.
This paper presents a spatiotemporal unsupervised feature learning method for cause identification of electromagnetic transient events (EMTE) in power grids. The proposed method is formulated based on the availability of time-synchronized high-frequency measurement, and using the convolutional neural network (CNN) as t…
A new algorithm difFOCI improves feature learning from data.
problem Feature selection and learning from data.
method Parametric, differentiable approximation of FOCI method.
result Improves feature learning with better management of spurious correlations.
Study links neural network inductive bias, feature learning, and generalization on Boolean functions.
problem Understanding how neural networks learn and generalize on Boolean data.
method End-to-end analysis of depth-2 discrete fully connected networks and DNF formulas, using Monte Carlo learning.
result Predictable training dynamics and interpretable features emerge, linking inductive bias and generalization.
Paper proposes a new method to learn features from error representations.
problem Learning from error representations in machine learning.
method Inverse feature learning (IFL) based on deep clustering.
result IFL leads to improved performance in classification and clustering.
Gradient descent reshapes the function space of neural networks.
problem Understanding how feature learning affects the function space of neural networks.
method Characterized the evolution of the feature space during training using a two-layer neural network.
result Gradient descent induces a data-adaptive deformation that selectively enhances signal-aligned directions.
Adaptive kernels from neural networks improve model performance.
problem Improving neural network performance through adaptive kernels.
method Deriving adaptive kernels from infinite-width neural networks using feature learning and gradient flow training.
result Adaptive kernels achieve lower test loss compared to traditional kernels.
This work explores feature learning tradeoffs in neural networks.
problem Resource tradeoffs in neural feature learning.
method Theoretical and experimental investigation of offline sparse parity learning.
result Width improves sample efficiency in sparse feature learning.
Three-layer networks learn more complex features than two-layer networks.
problem Understanding feature learning in deep neural networks.
method Analysis of three-layer neural networks trained with gradient descent.
result Three-layer networks can learn functions that two-layer networks cannot.
New insights into neural network feature learning through multi-step gradient descent.
problem Understanding feature learning in two-layer neural networks with limited width.
method Characterization of feature learning through two steps of gradient descent with specific step sizes.
result The second step of gradient descent reveals multiple learned directions, not limited to a single direction as in the first step.
A new method learns outcome-aware spectral features for causal effect estimation.
problem Estimation of causal effects in the presence of hidden confounders.
method Augmented Spectral Feature Learning framework that minimizes a contrastive loss derived from an augmented operator incorporating outcome information.
result Our method remains effective even under spectral misalignment.
Optimal feature learning strength improves generalization in deep networks.
problem Understanding how feature learning strength affects generalization in practical settings.
method Empirical studies and theoretical analysis of gradient flow dynamics in two-layer ReLU nets.
result Optimal feature learning strength yields substantial generalization gains, contrary to the prevailing intuition.
Due to the rapid innovation of technology and the desire to find and employ biomarkers for neurodegenerative disease, high-dimensional data classification problems are routinely encountered in neuroimaging studies. To avoid over-fitting and to explore relationships between disease and potential biomarkers, feature lear…
Breaking symmetry in training data is key for generalization in feature learning kernels.
problem Grokking in algebraic tasks, where models perform well on training but fail on unseen data.
method Used Recursive Feature Machine (RFM) with AGOP to learn task-relevant features, breaking symmetry in training data.
result Generalization occurs only when symmetry in the training set is broken, and RFM generalizes by recovering underlying invariance group action.
The paper explains spikes in training loss as catapults, improving feature learning and generalization.
problem Understanding and improving the training process of neural networks.
method Analysis of training loss spikes in SGD, empirical evidence of catapults, and demonstration of AGOP alignment.
result Catapults in training loss promote feature learning and better generalization.
Recently, considerable effort has been devoted to deep domain adaptation in computer vision and machine learning communities. However, most of existing work only concentrates on learning shared feature representation by minimizing the distribution discrepancy across different domains. Due to the fact that all the domai…
Sublinearly structured DNNs achieve feature learning consistency for compositional functions.
problem Achieving feature-learning and prediction consistency in deep neural networks.
method Sublinearly structured DNNs
result Sublinearly structured DNNs match or surpass wide DNNs in prediction.
ICA reveals deep learning's feature learning mechanisms from non-Gaussian data.
problem Understanding feature learning from non-Gaussian inputs in deep neural networks.
method Investigates ICA and SGD on synthetic and real data.
result FastICA requires n≳d4 samples for single non-Gaussian direction recovery, while SGD outperforms and optimised SGD reaches n≳d2. Mixup improves feature learning by mixing common and rare features.
problem Improving generalization in deep learning models.
method Mixup, a data augmentation technique, is applied to feature learning. Theoretical and experimental studies are conducted to understand its benefits.
result Mixup effectively learns rare features from common ones, leading to better generalization.
Deep networks learn features suddenly, akin to a phase transition.
problem Understanding sudden feature learning in deep networks.
method Adaptive kernel approach applied to two teacher-student models.
result Feature learning and Grokking are analogous to a first-order phase transition.
RFM reduces feature space for linear models, improving sparse recovery.
problem Sparse linear regression and low-rank matrix recovery.
method Recursive Feature Machines (RFM) that alternates between reweighting feature vectors by AGOP and learning prediction function.
result RFM generalizes IRLS and outperforms deep linear networks.
Kernel models learn low-dimensional predictive subspaces from input data.
problem Learning effective feature transformations in kernel models.
method Study of a compositional kernel ridge regression model.
result Global minimizers of the objective function identify the subspace with high probability.
Analyzes feature learning in neural networks using a self-consistent dynamical field theory.
problem Feature learning in infinite-width neural networks.
method Constructs deterministic dynamical order parameters as inner-product kernels for hidden unit activations and gradients.
result Reveals the hidden layer activation distribution, neural tangent kernel evolution, and output predictions.
Gradient descent aligns neural feature matrices with pre-activation tangent features.
problem Understanding neural feature learning mechanisms.
method Analytical proof of alignment between weight matrices and pre-activation tangent features.
result Derivative alignment occurs almost surely in high-dimensional settings.
Study on Bayesian deep linear networks with multiple outputs and convolutional layers.
problem Characterize feature learning in finite-width Bayesian deep linear networks.
method Exact and analytical formulas for joint and posterior distributions, using large deviation theory.
result Quantitative description of feature learning in infinite-width regime.
This paper proposes a practical approach for automatic sleep stage classification based on a multi-level feature learning framework and Recurrent Neural Network (RNN) classifier using heart rate and wrist actigraphy derived from a wearable device. The feature learning framework is designed to extract low- and mid-level…