The automation of the medical evidence acquisition and diagnosis process has recently attracted increasing attention in order to reduce the workload of doctors and democratize access to medical care. However, most works proposed in the machine learning literature focus solely on improving the prediction accuracy of a patient's pathology. We argue that this objective is insufficient to ensure doctors' acceptability of such systems. In their initial interaction with patients, doctors do not only focus on identifying the pathology a patient is suffering from; they instead generate a differential diagnosis (in the form of a short list of plausible diseases) because the medical evidence collected from patients is often insufficient to establish a final diagnosis. Moreover, doctors explicitly explore severe pathologies before potentially ruling them out from the differential, especially in acute care settings. Finally, for doctors to trust a system's recommendations, they need to understand how the gathered evidences led to the predicted diseases. In particular, interactions between a system and a patient need to emulate the reasoning of doctors. We therefore propose to model the evidence acquisition and automatic diagnosis tasks using a deep reinforcement learning framework that considers three essential aspects of a doctor's reasoning, namely generating a differential diagnosis using an exploration-confirmation approach while prioritizing severe pathologies. We propose metrics for evaluating interaction quality based on these three aspects. We show that our approach performs better than existing models while maintaining competitive pathology prediction accuracy.

There has been a rapidly growing interest in Automatic Symptom Detection (ASD) and Automatic Diagnosis (AD) systems in the machine learning research literature, aiming to assist doctors in telemedicine services. These systems are designed to interact with patients, collect evidence about their symptoms and relevant antecedents, and possibly make predictions about the underlying diseases. Doctors would review the interactions, including the evidence and the predictions, collect if necessary additional information from patients, before deciding on next steps. Despite recent progress in this area, an important piece of doctors' interactions with patients is missing in the design of these systems, namely the differential diagnosis. Its absence is largely due to the lack of datasets that include such information for models to train on. In this work, we present a large-scale synthetic dataset of roughly 1.3 million patients that includes a differential diagnosis, along with the ground truth pathology, symptoms and antecedents for each patient. Unlike existing datasets which only contain binary symptoms and antecedents, this dataset also contains categorical and multi-choice symptoms and antecedents useful for efficient data collection. Moreover, some symptoms are organized in a hierarchy, making it possible to design systems able to interact with patients in a logical way. As a proof-of-concept, we extend two existing AD and ASD systems to incorporate the differential diagnosis, and provide empirical evidence that using differentials as training signals is essential for the efficiency of such systems or for helping doctors better understand the reasoning of those systems.

Biasing properly the hypothesis space of a learner has been shown to improve generalization performance. Methods for achieving this goal have been proposed, that range from designing and introducing a bias into a learner to automatically learning the bias. Multitask learning methods fall into the latter category. When several related tasks derived from the same domain are available, these methods use the domain-related knowledge coded in the training examples of all the tasks as a source of bias. We extend some of the ideas presented in this field and describe a new approach that identifies a family of hypotheses, represented by a manifold in hypothesis space, that embodies domain-related knowledge. This family is learned using training examples sampled from a group of related tasks. Learning models trained on these tasks are only allowed to select hypotheses that belong to the family. We show that the new approach encompasses a large variety of families which can be learned. A statistical analysis on a class of related tasks is performed that shows significantly improved performances when using this approach.

Estimating insurance premia from data is a difficult regression problem for several reasons: the large number of variables, many of which are .discrete, and the very peculiar shape of the noise distri(cid:173) bution, asymmetric with fat tails, with a large majority zeros and a few unreliable and very large values. We compare several machine learning methods for estimating insurance premia, and test them on a large data base of car insurance policies. We find that func(cid:173) tion approximation methods that do not optimize a squared loss, like Support Vector Machines regression, do not work well in this context. Compared methods include decision trees and generalized linear models. The best results are obtained with a mixture of experts, which better identifies the least and most risky contracts, and allows to reduce the median premium by charging more to the most risky customers.

Biasing the hypothesis space of a learner has been shown to improve generalisation performances. Methods for achieving this goal are proposed, that range from deriving and introducing a bias into a learner to automatically learning the bias. In the latter case, most methods learn the bias by simultaneously training several related tasks derived from the same domain and imposing constraints on their parameters. We extend some of the ideas presented in this field and describe a new model that parametrizes the parameters of each task as a function of an affine manifold defined in parameter space and a point lying on the manifold. An analysis of variance on a class of learning tasks is performed that shows some significantly improved performances when using the model.

We consider the problem of feature-based face recognition in the setting where only a single example of each face is available for training. The mixture-distance technique we introduce achieves a recognition rate of 95% on a database of 685 people in which each face is represented by 30 measured distances. This is currently the best recorded recognition rate for a feature-based system applied to a database of this size. By comparison, nearest neighbor search using Euclidean distance yields 84%. In our work a novel distance function is constructed based on local second order statistics as estimated by modeling the training data as a mixture of normal densities. We report on the results from mixtures of several sizes. We demonstrate that a flat mixture of mixtures performs as well as the best model and therefore represents an effective solution to the model selection problem. A mixture perspective is also taken for individual Gaussians to choose between first order (variance) and second order (covariance) models. Here an approximation to flat combination is proposed and seen to perform well in practice. Our results demonstrate that even in the absence of multiple training examples for each class, it is sometimes possible to infer from a statistical model of training data, a significantly improved distance function for use in pattern recognition.