Skip to Main content Skip to Navigation
Journal articles

Compression-Based Regularization with an Application to Multi-Task Learning

Abstract : This paper investigates, from information theoretic grounds, a learning problem based on the principle that any regularity in a given dataset can be exploited to extract compact features from data, i.e., using fewer bits than needed to fully describe the data itself, in order to build meaningful representations of a relevant content (multiple labels). We begin by introducing the noisy lossy source coding paradigm with the log-loss fidelity criterion which provides the fundamental tradeoffs between the cross-entropy loss (average risk) and the information rate of the features (model complexity). Our approach allows an information theoretic formulation of the multi-task learning (MTL) problem which is a supervised learning framework in which the prediction models for several related tasks are learned jointly from common representations to achieve better generalization performance. Then, we present an iterative algorithm for computing the optimal tradeoffs and its global convergence is proven provided that some conditions hold. An important property of this algorithm is that it provides a natural safeguard against overfitting, because it minimizes the average risk taking into account a penalization induced by the model complexity. Remarkably, empirical results illustrate that there exists an optimal information rate minimizing the excess risk which depends on the nature and the amount of available training data. An application to hierarchical text categorization is also investigated, extending previous works.
Document type :
Journal articles
Complete list of metadatas

Cited literature [42 references]  Display  Hide  Download

https://hal-centralesupelec.archives-ouvertes.fr/hal-02944069
Contributor : Delphine Le Piolet <>
Submitted on : Monday, September 21, 2020 - 10:49:09 AM
Last modification on : Monday, September 28, 2020 - 3:59:50 PM

File

compression_based_regularizati...
Files produced by the author(s)

Identifiers

Citation

Matias Vera, Leonardo Rey Vega, Pablo Piantanida. Compression-Based Regularization with an Application to Multi-Task Learning. IEEE Journal of Selected Topics in Signal Processing, IEEE, 2018, 12 (5), pp.1063-1076. ⟨10.1109/JSTSP.2018.2846218⟩. ⟨hal-02944069⟩

Share

Metrics

Record views

14

Files downloads

33