Paper studies early-stopped mirror descent for noisy sparse phase retrieval.
problem Recovering a sparse signal from noisy quadratic measurements.
method Early-stopped mirror descent with hyperbolic entropy mirror map.
result Achieves nearly minimax-optimal rate of convergence for k-sparse signals. Early training phase affects deep neural network optimization and generalization.
problem The choice of learning rate influences generalization in deep learning models.
method Showed that SGD implicitly penalizes the trace of the Fisher Information Matrix (FIM) from the start of training, and explicitly penalizing the trace of FIM improves generalization.
result Catastrophic Fisher explosion (large trace of FIM early in training) is linked to poor generalization.
Early SGD hyperparameters affect deep neural network training, showing a break-even point.
problem Understanding how early SGD hyperparameters influence deep neural network training.
method Analysis of stochastic gradient descent (SGD) hyperparameters and their effects on the optimization trajectory.
result A break-even point exists where SGD implicitly regularizes the loss surface and improves gradient conditioning.
Framework prevents deep learning models from memorizing noisy labels.
problem Deep learning models memorize noisy labels during early learning phase.
method Develops a technique that exploits early learning phase via regularization.
result Framework achieves robustness to noisy annotations on benchmarks and real-world datasets.
Early stopping improves generalization in overparameterized diffusion models.
problem Understanding and optimizing generalization in overparameterized diffusion models.
method Revisiting diffusion models, showing generalization occurs before memorization, and developing a phase diagram.
result Generalization time scales with dataset size, supporting early-stopping criteria.
Early neural network training reveals important sub-networks and weight distributions.
problem Understanding the early phases of neural network training.
method Extensive measurements and quantitative probing of weight distribution and dataset reliance.
result Deep networks are not robust to reinitializing with random weights while maintaining signs, and weight distributions are highly non-independent.
This research examines how the error rate of nearest neighbor classifiers varies with dataset size.
problem The scaling of classification error rates with dataset size is not uniform.
method Theoretical analysis of nearest neighbor classifiers, focusing on early and late phases of dataset size.
result The error rate of nearest neighbor classifiers can have fine-grained rates depending on the dataset size and data distribution.
FL's early training phase significantly impacts final test accuracy.
problem Understanding how early phases affect FL's final test accuracy.
method Generalized Fisher Information Matrix (FedFIM) to FL.
result FL exhibits critical learning periods where small errors can have large impacts.
The paper develops a method to predict the latent deterioration phase in limit order books before stress is observed.
problem Limit order books can transition rapidly from stable to stressed conditions, making it difficult to detect the latent deterioration phase.
method The paper formalizes a three-regime causal data-generating process and proposes a trigger-based detector combining MAX aggregation of complementary signal channels, a rising-edge condition, and adaptive thresholding.
result The proposed method achieves mean lead-time of +18.6 timesteps with perfect precision and moderate coverage, outperforming classical change-point and microstructure baselines.
Study uses machine learning to detect early COVID-19 from CT images.
problem Early detection of COVID-19 from CT images.
method Machine learning methods applied to patches of CT images, feature extraction (GLCM, LDP, GLRLM, GLSZM, DWT), SVM classification.
result Best classification accuracy of 99.68% with 10-fold cross-validation and GLSZM feature extraction.
Noisy labels are very common in real-world training data, which lead to poor generalization on test data because of overfitting to the noisy labels. In this paper, we claim that such overfitting can be avoided by "early stopping" training a deep neural network before the noisy labels are severely memorized. Then, we re…
The paper models market crashes as phase transitions, finding dynamic transitions offer better predictions.
problem Understanding and predicting extreme financial events like market crashes.
method Employing phase transition theory, focusing on endogenous crashes, and comparing DPT, CPT, and SPT.
result Dynamic phase transitions provide more accurate predictions of market crashes compared to critical and stochastic models.
CDLF predicts product life-cycles in cold-start phases with high accuracy.
problem Forecasting new products in early phases when data is scarce.
method Conditional Diffusion Life-cycle Forecaster (CDLF) combining static descriptors, reference trajectories, and new observations.
result CDLF outperforms classical models in accuracy and probabilistic forecasting.
Catapult phase in neural nets shows exponential loss growth before quick decrease.
problem Understanding phase transitions in neural networks during training.
method Analyzing weight norm and loss behavior for super-critical learning rates.
result Proven existence of catapult phase in quadratic models and two-layer nets.
Theoretical analysis of t-SNE for visualizing clustered data.
problem Understanding t-SNE for visualizing high-dimensional clustered data.
method Gradient descent approach and power iterations based on graph Laplacian.
result Asymptotic equivalence and limiting behavior of t-SNE's early exaggeration stage.
Dynamic Influence Tracker measures changing sample importance during model training.
problem Static influence measurements during training overlook how sample importance varies over time.
method Dynamic Influence Tracker (DIT) captures time-varying sample influence across arbitrary time windows.
result DIT reveals distinct learning phases with shifting priorities and detects corrupted samples more efficiently.
Minimizing non-convex and high-dimensional objective functions is challenging, especially when training modern deep neural networks. In this paper, a novel approach is proposed which divides the training process into two consecutive phases to obtain better generalization performance: Bayesian sampling and stochastic op…
Early neural networks can be simplified to linear models, revealing surprising simplicity.
problem Complexity of neural network learning dynamics.
method Formal proof and empirical verification of early-time learning dynamics of neural networks.
result Early learning dynamics of neural networks can be approximated by simple linear models.
This study uses high-frequency data to identify early warning signals for bank crises.
problem Identifying early warning signals for impending bank crises.
method Constructing multiple recurrence networks (MRNs) based on high-frequency stock returns to monitor nonlinear dynamics.
result Key indicators of MRNs, particularly average mutual information, provide valuable insights into periods of extreme volatility.
Weyl and Cartan proposed different but related ways to handle infinitesimal geometry in the early 1920s.
problem How to apply transformation groups in differential geometry.
method Both used connections and parallel transfer, with Cartan aiming for a more general framework.
result They reached an agreement on handling Cartan's infinitesimal geometric structures by the 1930s.
Early alignment in neural networks leads to sparse representations but hinders convergence.
problem The implicit bias of gradient descent during early training phases.
method Quantitative description of early alignment phase in small initialisation, one hidden layer networks.
result Early alignment induces a sparse representation but also hinders convergence to global minima.
SpecGD mitigates misalignment in phase retrieval models with anisotropic inputs.
problem Misalignment during gradient descent in phase retrieval models with anisotropic inputs.
method Spectral gradient descent modifies gradient updates to preserve directional information and remove spike amplification.
result SpecGD removes spike amplification, leading to stable alignment and accelerated noise contraction.
We present here an introduction to Brainstorming approach, that was recently proposed as a consensus meta-learning technique, and used in several practical applications in bioinformatics and chemoinformatics. The consensus learning denotes heterogeneous theoretical classification method, where one trains an ensemble of…
Since 2007, several contributions have tried to identify early-warning signals of the financial crisis. However, the vast majority of analyses has focused on financial systems and little theoretical work has been done on the economic counterpart. In the present paper we fill this gap and employ the theoretical tools of…
What enables Stochastic Gradient Descent (SGD) to achieve better generalization than Gradient Descent (GD) in Neural Network training? This question has attracted much attention. In this paper, we study the distribution of the Stochastic Gradient Noise (SGN) vectors during the training. We observe that for batch sizes …
A novel framework IMBoost improves outlier detection by leveraging the inlier memorization effect.
problem Challenges in unsupervised outlier detection, especially when inliers and outliers are not well-separated or form dense clusters.
method IMBoost framework that incorporates active learning to selectively acquire informative labels and explicitly reinforce the inlier memorization effect.
result IMBoost significantly outperforms state-of-the-art active outlier detection methods and requires less computational cost.
We study the problem of finding the optimal dosage in early stage clinical trials through the multi-armed bandit lens. We advocate the use of the Thompson Sampling principle, a flexible algorithm that can accommodate different types of monotonicity assumptions on the toxicity and efficacy of the doses. For the simplest…
We present RAPID (Real-time Automated Photometric IDentification), a novel time-series classification tool capable of automatically identifying transients from within a day of the initial alert, to the full lifetime of a light curve. Using a deep recurrent neural network with Gated Recurrent Units (GRUs), we present th…
Two-layer ReLU networks often converge to simpler solutions, improving generalization.
problem Understanding generalization in overparametrized neural networks, especially for complex tasks.
method Theoretical analysis of two-layer ReLU networks, focusing on the early alignment phase.
result Two-layer ReLU networks often converge to simpler solutions rather than interpolating the training data, leading to better generalization.
Random matrix theory explains transient signal detectability in early-stopped gradient flow.
problem Transient signal detectability in early-stopped gradient flow.
method Random matrix theory applied to gradient flow in a linear teacher-student setting.
result Transient Baik-Ben Arous-Péché (BBP) transition in learning dynamics due to anisotropy and noise.
This paper simplifies fine-tuning for small LLMs, reducing barriers for developers.
problem Limited resources for fine-tuning large language models (LLMs) by individual developers and small organizations.
method Instruction-tuning datasets, small-sized LLMs (3B to 7B parameters), various training configurations and strategies.
result Improved model performance on benchmarks with specific training configurations, and insights into early termination and hyperparameter simplifications.
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.
Study shows deep linear networks can converge to flatter minima at large learning rates.
problem Understanding the implicit bias of deep linear networks at large learning rates.
method Characterization of deep linear networks for binary classification using logistic loss in the large learning rate regime.
result Gradient descent iterates converge to a flatter minimum in the catapult phase for certain data separation conditions.
Lottery tickets find good initializations for IMP with sparse training.
problem Finding good initializations for iterative magnitude pruning (IMP) in sparse networks.
method Empirical study of IMP performance with varying pre-training data and iterations.
result Training on a small fraction of data suffices to obtain good initializations for IMP.
Gradient descent dynamics in neural networks show quenching and activation phases.
problem Understanding training dynamics in neural networks.
method Numerical and phenomenological study of gradient descent algorithm for two-layer neural networks.
result Gradient descent dynamics exhibit quenching and activation phases in under-parametrized networks.
A new control chart detects shifts in binary data streams quickly and reliably.
problem Early detection of small shifts in multiple binary data streams.
method Cumulative Standardized Binomial EWMA (CSB-EWMA) chart with exact variance derivation.
result Adaptive control limits ensure robust detection across different data distributions.
We present a feature engineering pipeline for the construction of musical signal characteristics, to be used for the design of a supervised model for musical genre identification. The key idea is to extend the traditional two-step process of extraction and classification with additive stand-alone phases which are no lo…
Theoretical study shows AI models can recover from contaminated training data.
problem Data contamination in AI training can degrade model performance.
method Theoretical analysis and experiments on various data types.
result Models converge to true distribution under mild conditions, with rate dependent on real data fraction.
A new era in radioastronomy will begin with the upcoming large-scale surveys planned at the Australian Square Kilometre Array Pathfinder (ASKAP). ASKAP started its Early Science program in October 2017 and several target fields were observed during the array commissioning phase. The SCORPIO field was the first observed…
Margin enlargement over training data has been an important strategy since perceptrons in machine learning for the purpose of boosting the robustness of classifiers toward a good generalization ability. Yet Breiman (1999) showed a dilemma that a uniform improvement on margin distribution does NOT necessarily reduces ge…
Machine learning predicts failure in brittle materials with high accuracy.
problem Predicting failure in brittle materials under repetitive loads.
method Phase-field model combined with supervised machine learning.
result Framework predicts failure with acceptable accuracy even in noisy data.
SAM selects flatter minima late in training, improving generalization.
problem Improving neural network generalization under various settings.
method Sharpness-Aware Minimization (SAM) applied late in training.
result SAM efficiently selects flatter minima late in training, improving generalization.
The evolution with time of the correlation structure of equity returns is studied by means of a filtered network approach investigating persistences and recurrences and their implications for risk diversification strategies. We build dynamically Planar Maximally Filtered Graphs from the correlation structure over a rol…
Every year, 3 million newborns die within the first month of life. Birth asphyxia and other breathing-related conditions are a leading cause of mortality during the neonatal phase. Current diagnostic methods are too sophisticated in terms of equipment, required expertise, and general logistics. Consequently, early dete…
Early approaches to multiple-output Gaussian processes (MOGPs) relied on linear combinations of independent, latent, single-output Gaussian processes (GPs). This resulted in cross-covariance functions with limited parametric interpretation, thus conflicting with the ability of single-output GPs to understand lengthscal…
In the early phases of the product life cycle, the costs controls became a major decision tool in the competitiveness of the companies due to the world competition. After defining the problems related to this control difficulties, we will present an approach using a concept of cost entity related to the design and real…
SAPAG attacks distributed learning by reconstructing true training data from gradients.
problem Privacy attacks on distributed learning systems through gradients.
method SAPAG uses a Gaussian kernel-based gradient difference distance measure.
result SAPAG can reconstruct training data on various DNNs and at different training phases.
Study uses LSTM models to detect Wyckoff patterns in currency trading.
problem Understanding market dynamics and identifying trading opportunities.
method Dissecting Wyckoff Phases, using CNNs for spatial data and LSTM for temporal data.
result Deep learning models enhance pattern recognition in financial markets.