Pascal Germain

Domain-adversarial training of neural networks

We introduce a new representation learning approach for domain adaptation, in which data at training and test time come from similar but different distributions. Our approach is directly inspired by the theory on domain adaptation suggesting that, for effective domain transfer to be achieved, predictions must be made based on features that cannot discriminate between the training (source) and test (target) domains. The approach implements this idea in the context of neural network architectures that are trained on labeled data from the source domain and unlabeled data from the target domain (no labeled target-domain data is necessary). As the training progresses, the approach promotes the emergence of features that are (i) discriminative for the main learning task on the source domain and (ii) indiscriminate with respect to the shift between the domains. We show that this adaptation behaviour can be achieved in almost any feed-forward model by augmenting it with few standard layers and a new gradient reversal layer. The resulting augmented architecture can be trained using standard backpropagation and stochastic gradient descent, and can thus be implemented with little effort using any of the deep learning packages. We demonstrate the success of our approach for two distinct classification problems (document sentiment analysis and image classification), where state-of-the-art domain adaptation performance on standard benchmarks is achieved. We also validate the approach for descriptor learning task in the context of person re-identification application.

PAC-Bayesian learning of linear classifiers

We present a general PAC-Bayes theorem from which all known PAC-Bayes risk bounds are obtained as particular cases. We also propose different learning algorithms for finding linear classifiers that minimize these bounds. These learning algorithms are generally competitive with both AdaBoost and the SVM.

A pseudo-boolean set covering machine

The Set Covering Machine (SCM) is a machine learning algorithm that constructs a conjunction of Boolean functions. This algorithm is motivated by the minimization of a theoretical bound. However, finding the optimal conjunction according to this bound is a combinatorial problem. The SCM approximates the solution using a greedy approach. Even though SCM seems very efficient in practice, it is unknown how it compares to the optimal solution. To answer this question, we present a novel pseudo-Boolean optimization model that encodes the minimization problem. It is the first time a Constraint Programming approach addresses the combinatorial problem related to this machine learning algorithm. Using that model and recent pseudo-Boolean solvers, we empirically show that the greedy approach is surprisingly close to the optimal.