Despite the popularity and success of deep learning, there is limited understanding of when, how, and why neural networks generalize to unseen examples. Since learning can be seen as extracting information from data, we formally study information captured by neural networks during training. Specifically, we start with viewing learning in presence of noisy labels from an information-theoretic perspective and derive a learning algorithm that limits label noise information in weights. We then define a notion of unique information that an individual sample provides to the training of a deep network, shedding some light on the behavior of neural networks on examples that are atypical, ambiguous, or belong to underrepresented subpopulations. We relate example informativeness to generalization by deriving nonvacuous generalization gap bounds. Finally, by studying knowledge distillation, we highlight the important role of data and label complexity in generalization. Overall, our findings contribute to a deeper understanding of the mechanisms underlying neural network generalization.
Neural networks employ spurious correlations in their predictions, resulting in decreased performance when these correlations do not hold. Recent works suggest fixing pretrained representations and training a classification head that does not use spurious features. We investigate how spurious features are represented in pretrained representations and explore strategies for removing information about spurious features. Considering the Waterbirds dataset and a few pretrained representations, we find that even with full knowledge of spurious features, their removal is not straightforward due to entangled representation. To address this, we propose a linear autoencoder training method to separate the representation into core, spurious, and other features. We propose two effective spurious feature removal approaches that are applied to the encoding and significantly improve classification performance measured by worst group accuracy.
We propose an approach to estimate the number of samples required for a model to reach a target performance. We find that the power law, the de facto principle to estimate model performance, leads to large error when using a small dataset (e.g., 5 samples per class) for extrapolation. This is because the log-performance error against the log-dataset size follows a nonlinear progression in the few-shot regime followed by a linear progression in the high-shot regime. We introduce a novel piecewise power law (PPL) that handles the two data regimes differently. To estimate the parameters of the PPL, we introduce a random forest regressor trained via meta learning that generalizes across classification/detection tasks, ResNet/ViT based architectures, and random/pre-trained initializations. The PPL improves the performance estimation on average by 37% across 16 classification and 33% across 10 detection datasets, compared to the power law. We further extend the PPL to provide a confidence bound and use it to limit the prediction horizon that reduces over-estimation of data by 76% on classification and 91% on detection datasets.
Despite the popularity and efficacy of knowledge distillation, there is limited understanding of why it helps. In order to study the generalization behavior of a distilled student, we propose a new theoretical framework that leverages supervision complexity: a measure of alignment between teacher-provided supervision and the student's neural tangent kernel. The framework highlights a delicate interplay among the teacher's accuracy, the student's margin with respect to the teacher predictions, and the complexity of the teacher predictions. Specifically, it provides a rigorous justification for the utility of various techniques that are prevalent in the context of distillation, such as early stopping and temperature scaling. Our analysis further suggests the use of online distillation, where a student receives increasingly more complex supervision from teachers in different stages of their training. We demonstrate efficacy of online distillation and validate the theoretical findings on a range of image classification benchmarks and model architectures.
Some of the tightest information-theoretic generalization bounds depend on the average information between the learned hypothesis and a \emph{single} training example. However, these sample-wise bounds were derived only for \emph{expected} generalization gap. We show that even for expected \emph{squared} generalization gap no such sample-wise information-theoretic bounds exist. The same is true for PAC-Bayes and single-draw bounds. Remarkably, PAC-Bayes, single-draw and expected squared generalization gap bounds that depend on information in pairs of examples exist.
Domain generalization algorithms use training data from multiple domains to learn models that generalize well to unseen domains. While recently proposed benchmarks demonstrate that most of the existing algorithms do not outperform simple baselines, the established evaluation methods fail to expose the impact of various factors that contribute to the poor performance. In this paper we propose an evaluation framework for domain generalization algorithms that allows decomposition of the error into components capturing distinct aspects of generalization. Inspired by the prevalence of algorithms based on the idea of domain-invariant representation learning, we extend the evaluation framework to capture various types of failures in achieving invariance. We show that the largest contributor to the generalization error varies across methods, datasets, regularization strengths and even training lengths. We observe two problems associated with the strategy of learning domain-invariant representations. On Colored MNIST, most domain generalization algorithms fail because they reach domain-invariance only on the training domains. On Camelyon-17, domain-invariance degrades the quality of representations on unseen domains. We hypothesize that focusing instead on tuning the classifier on top of a rich representation can be a promising direction.
We derive information-theoretic generalization bounds for supervised learning algorithms based on the information contained in predictions rather than in the output of the training algorithm. These bounds improve over the existing information-theoretic bounds, are applicable to a wider range of algorithms, and solve two key challenges: (a) they give meaningful results for deterministic algorithms and (b) they are significantly easier to estimate. We show experimentally that the proposed bounds closely follow the generalization gap in practical scenarios for deep learning.
We define a notion of information that an individual sample provides to the training of a neural network, and we specialize it to measure both how much a sample informs the final weights and how much it informs the function computed by the weights. Though related, we show that these quantities have a qualitatively different behavior. We give efficient approximations of these quantities using a linearized network and demonstrate empirically that the approximation is accurate for real-world architectures, such as pre-trained ResNets. We apply these measures to several problems, such as dataset summarization, analysis of under-sampled classes, comparison of informativeness of different data sources, and detection of adversarial and corrupted examples. Our work generalizes existing frameworks but enjoys better computational properties for heavily over-parametrized models, which makes it possible to apply it to real-world networks.
In the presence of noisy or incorrect labels, neural networks have the undesirable tendency to memorize information about the noise. Standard regularization techniques such as dropout, weight decay or data augmentation sometimes help, but do not prevent this behavior. If one considers neural network weights as random variables that depend on the data and stochasticity of training, the amount of memorized information can be quantified with the Shannon mutual information between weights and the vector of all training labels given inputs, $I(w : \mathbf{y} \mid \mathbf{x})$. We show that for any training algorithm, low values of this term correspond to reduction in memorization of label-noise and better generalization bounds. To obtain these low values, we propose training algorithms that employ an auxiliary network that predicts gradients in the final layers of a classifier without accessing labels. We illustrate the effectiveness of our approach on versions of MNIST, CIFAR-10, and CIFAR-100 corrupted with various noise models, and on a large-scale dataset Clothing1M that has noisy labels.
Estimating the covariance structure of multivariate time series is a fundamental problem with a wide-range of real-world applications -- from financial modeling to fMRI analysis. Despite significant recent advances, current state-of-the-art methods are still severely limited in terms of scalability, and do not work well in high-dimensional undersampled regimes. In this work we propose a novel method called Temporal Correlation Explanation, or T-CorEx, that (a) has linear time and memory complexity with respect to the number of variables, and can scale to very large temporal datasets that are not tractable with existing methods; (b) gives state-of-the-art results in highly undersampled regimes on both synthetic and real-world datasets; and (c) makes minimal assumptions about the character of the dynamics of the system. T-CorEx optimizes an information-theoretic objective function to learn a latent factor graphical model for each time period and applies two regularization techniques to induce temporal consistency of estimates. We perform extensive evaluation of T-Corex using both synthetic and real-world data and demonstrate that it can be used for detecting sudden changes in the underlying covariance matrix, capturing transient correlations and analyzing extremely high-dimensional complex multivariate time series such as high-resolution fMRI data.