اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدEvading the Simplicity Bias: Training a Diverse Set of Models Discovers Solutions with Superior OOD Generalization
72
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Neural networks trained with SGD were recently shown to rely preferentially on linearly-predictive features and can ignore complex, equally-predictive ones. This simplicity bias can explain their lack of robustness out of distribution (OOD). The more complex the task to learn, the more likely it is that statistical artifacts (i.e. selection biases, spurious correlations) are simpler than the mechanisms to learn. We demonstrate that the simplicity bias can be mitigated and OOD generalization improved. We train a set of similar models to fit the data in different ways using a penalty on the alignment of their input gradients. We show theoretically and empirically that this induces the learning of more complex predictive patterns. OOD generalization fundamentally requires information beyond i.i.d. examples, such as multiple training environments, counterfactual examples, or other side information. Our approach shows that we can defer this requirement to an independent model selection stage. We obtain SOTA results in visual recognition on biased data and generalization across visual domains. The method - the first to evade the simplicity bias - highlights the need for a better understanding and control of inductive biases in deep learning.
قيم البحث
اقرأ أيضاً
Recently, learning a model that generalizes well on out-of-distribution (OOD) data has attracted great attention in the machine learning community. In this paper, after defining OOD generalization via Wasserstein distance, we theoretically show that
a model robust to input perturbation generalizes well on OOD data. Inspired by previous findings that adversarial training helps improve input-robustness, we theoretically show that adversarially trained models have converged excess risk on OOD data, and empirically verify it on both image classification and natural language understanding tasks. Besides, in the paradigm of first pre-training and then fine-tuning, we theoretically show that a pre-trained model that is more robust to input perturbation provides a better initialization for generalization on downstream OOD data. Empirically, after fine-tuning, this better-initialized model from adversarial pre-training also has better OOD generalization.
Modern deep neural networks are highly over-parameterized compared to the data on which they are trained, yet they often generalize remarkably well. A flurry of recent work has asked: why do deep networks not overfit to their training data? We invest
igate the hypothesis that deeper nets are implicitly biased to find lower rank solutions and that these are the solutions that generalize well. We prove for the asymptotic case that the percent volume of low effective-rank solutions increases monotonically as linear neural networks are made deeper. We then show empirically that our claim holds true on finite width models. We further empirically find that a similar result holds for non-linear networks: deeper non-linear networks learn a feature space whose kernel has a lower rank. We further demonstrate how linear over-parameterization of deep non-linear models can be used to induce low-rank bias, improving generalization performance without changing the effective model capacity. We evaluate on various model architectures and demonstrate that linearly over-parameterized models outperform existing baselines on image classification tasks, including ImageNet.
Adversarially trained models exhibit a large generalization gap: they can interpolate the training set even for large perturbation radii, but at the cost of large test error on clean samples. To investigate this gap, we decompose the test risk into i
ts bias and variance components and study their behavior as a function of adversarial training perturbation radii ($varepsilon$). We find that the bias increases monotonically with $varepsilon$ and is the dominant term in the risk. Meanwhile, the variance is unimodal as a function of $varepsilon$, peaking near the interpolation threshold for the training set. This characteristic behavior occurs robustly across different datasets and also for other robust training procedures such as randomized smoothing. It thus provides a test for proposed explanations of the generalization gap. We find that some existing explanations fail this test--for instance, by predicting a monotonically increasing variance curve. This underscores the power of bias-variance decompositions in modern settings-by providing two measurements instead of one, they can rule out more explanations than test accuracy alone. We also show that bias and variance can provide useful guidance for scalably reducing the generalization gap, highlighting pre-training and unlabeled data as promising routes.
In order to train robust deep learning models, large amounts of labelled data is required. However, in the absence of such large repositories of labelled data, unlabeled data can be exploited for the same. Semi-Supervised learning aims to utilize suc
h unlabeled data for training classification models. Recent progress of self-training based approaches have shown promise in this area, which leads to this study where we utilize an ensemble approach for the same. A by-product of any semi-supervised approach may be loss of calibration of the trained model especially in scenarios where unlabeled data may contain out-of-distribution samples, which leads to this investigation on how to adapt to such effects. Our proposed algorithm carefully avoids common pitfalls in utilizing unlabeled data and leads to a more accurate and calibrated supervised model compared to vanilla self-training based student-teacher algorithms. We perform several experiments on the popular STL-10 database followed by an extensive analysis of our approach and study its effects on model accuracy and calibration.
Several works have proposed Simplicity Bias (SB)---the tendency of standard training procedures such as Stochastic Gradient Descent (SGD) to find simple models---to justify why neural networks generalize well [Arpit et al. 2017, Nakkiran et al. 2019,
Soudry et al. 2018]. However, the precise notion of simplicity remains vague. Furthermore, previous settings that use SB to theoretically justify why neural networks generalize well do not simultaneously capture the non-robustness of neural networks---a widely observed phenomenon in practice [Goodfellow et al. 2014, Jo and Bengio 2017]. We attempt to reconcile SB and the superior standard generalization of neural networks with the non-robustness observed in practice by designing datasets that (a) incorporate a precise notion of simplicity, (b) comprise multiple predictive features with varying levels of simplicity, and (c) capture the non-robustness of neural networks trained on real data. Through theory and empirics on these datasets, we make four observations: (i) SB of SGD and variants can be extreme: neural networks can exclusively rely on the simplest feature and remain invariant to all predictive complex features. (ii) The extreme aspect of SB could explain why seemingly benign distribution shifts and small adversarial perturbations significantly degrade model performance. (iii) Contrary to conventional wisdom, SB can also hurt generalization on the same data distribution, as SB persists even when the simplest feature has less predictive power than the more complex features. (iv) Common approaches to improve generalization and robustness---ensembles and adversarial training---can fail in mitigating SB and its pitfalls. Given the role of SB in training neural networks, we hope that the proposed datasets and methods serve as an effective testbed to evaluate novel algorithmic approaches aimed at avoiding the pitfalls of SB.
الأسئلة المقترحة
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
