Does learning of task-relevant representations stop when behavior stops changing? Motivated by recent theoretical advances in machine learning and the intuitive observation that human experts continue to learn from practice even after mastery, we hypothesize that task-specific representation learning can continue, even when behavior plateaus. In a novel reanalysis of recently published neural data, we find evidence for such learning in posterior piriform cortex of mice following continued training on a task, long after behavior saturates at near-ceiling performance ("overtraining"). This learning is marked by an increase in decoding accuracy from piriform neural populations and improved performance on held-out generalization tests. We demonstrate that class representations in cortex continue to separate during overtraining, so that examples that were incorrectly classified at the beginning of overtraining can abruptly be correctly classified later on, despite no changes in behavior during that time. We hypothesize this hidden yet rich learning takes the form of approximate margin maximization; we validate this and other predictions in the neural data, as well as build and interpret a simple synthetic model that recapitulates these phenomena. We conclude by showing how this model of late-time feature learning implies an explanation for the empirical puzzle of overtraining reversal in animal learning, where task-specific representations are more robust to particular task changes because the learned features can be reused.

A statistical theory for overtraining is proposed. The analysis treats realizable stochastic neural networks, trained with Kullback(cid:173) Leibler loss in the asymptotic case. It is shown that the asymptotic gain in the generalization error is small if we perform early stop(cid:173) ping, even if we have access to the optimal stopping time. Consider(cid:173) ing cross-validation stopping we answer the question: In what ratio the examples should be divided into training and testing sets in or(cid:173) der to obtain the optimum performance. In the non-asymptotic region cross-validated early stopping always decreases the general(cid:173) ization error.

This post was inspired by a recent post by Andrew Gelman, who defined'overfitting' as follows: Overfitting is when you have a complicated model that gives worse predictions, on average, than a simpler model. Preamble There is a lot of confusion among practitioners regarding the concept of overfitting. Applying cross-validation prevents overfitting and a good out-of-sample performance, low generalisation error in unseen data, indicates not an overfit. This statement is of course not true: cross-validation does not prevent your model to overfit and good out-of-sample performance does not guarantee not-overfitted model. What actually people refer to in one aspect of this statement is called overtraining.

A statistical theory for overtraining is proposed. The analysis treats realizable stochastic neural networks, trained with Kullback Leibler loss in the asymptotic case. It is shown that the asymptotic gain in the generalization error is small if we perform early stopping, even if we have access to the optimal stopping time. Considering cross-validation stopping we answer the question: In what ratio the examples should be divided into training and testing sets in order to obtain the optimum performance. In the non-asymptotic region cross-validated early stopping always decreases the generalization error. Our large scale simulations done on a CM5 are in nice agreement with our analytical findings.

A statistical theory for overtraining is proposed. The analysis treats realizable stochastic neural networks, trained with Kullback Leibler loss in the asymptotic case. It is shown that the asymptotic gain in the generalization error is small if we perform early stopping, even if we have access to the optimal stopping time. Considering cross-validation stopping we answer the question: In what ratio the examples should be divided into training and testing sets in order to obtain the optimum performance. In the non-asymptotic region cross-validated early stopping always decreases the generalization error. Our large scale simulations done on a CM5 are in nice agreement with our analytical findings.

A statistical theory for overtraining is proposed. The analysis treats realizable stochastic neural networks, trained with Kullback Leibler loss in the asymptotic case. It is shown that the asymptotic gain in the generalization error is small if we perform early stopping, evenif we have access to the optimal stopping time. Considering cross-validation stopping we answer the question: In what ratio the examples should be divided into training and testing sets in order toobtain the optimum performance. In the non-asymptotic region cross-validated early stopping always decreases the generalization error.Our large scale simulations done on a CM5 are in nice agreement with our analytical findings.