We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

stat.ML

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Statistics > Machine Learning

Title: Path classification by stochastic linear recurrent neural networks

Abstract: We investigate the functioning of a classifying biological neural network from the perspective of statistical learning theory, modelled, in a simplified setting, as a continuous-time stochastic recurrent neural network (RNN) with identity activation function. In the purely stochastic (robust) regime, we give a generalisation error bound that holds with high probability, thus showing that the empirical risk minimiser is the best-in-class hypothesis. We show that RNNs retain a partial signature of the paths they are fed as the unique information exploited for training and classification tasks. We argue that these RNNs are easy to train and robust and back these observations with numerical experiments on both synthetic and real data. We also exhibit a trade-off phenomenon between accuracy and robustness.
Subjects: Machine Learning (stat.ML); Information Theory (cs.IT); Machine Learning (cs.LG); Probability (math.PR)
MSC classes: 68T07, 68Q32 (Primary) 92B20, 60L10 (Secondary)
Cite as: arXiv:2108.03090 [stat.ML]
  (or arXiv:2108.03090v2 [stat.ML] for this version)

Submission history

From: Youness Boutaib [view email]
[v1] Fri, 6 Aug 2021 12:59:12 GMT (1126kb,D)
[v2] Fri, 7 Jan 2022 11:43:03 GMT (974kb,D)

Link back to: arXiv, form interface, contact.