Analyse 1,2,3,4
Algèbre 1,2,3
Probabilité 1,2
Recherche opérationnelle
Machine learning refers to a broad set of algorithms and related concerns for discovering patterns
in data, making new inferences based on data, and generally improving the performance of a
software system without direct programming. These methods are critical for data science. Data
scientists should understand the algorithms they apply, be able to implement them, if necessary,
and make principled decisions about their use.
Scope:
Broad categories of machine learning approaches (e.g.,supervised and unsupervised)
Algorithms and tools (i.e.,implementations of those algorithms) in each of the broad learning categories.
Problems related to model expressivity as well as availability of data, and techniques
Express formally the representational power of models learned by an algorithm, and relate that to issues such as expressiveness and overfitting.
Exhibit knowledge of methods to mitigate the effects of overfitting and curse of dimensionality in the context of machine learning algorithms.
Provide an appropriate performance metric for evaluating machine learning algorithms/tools for a given problem.
Differences in interpretability of learned models.
Solve the problem of overfitting, and unbalanced datasets.
Consultez les ressources disponibles concernant ce module sur le moteur de recherche de la bibliothèque, ou accédez directement au cours de vos enseignants via la plateforme de téléenseignement de l’école « e-learn ».