A mathematical model for automatic differentiation in machine learning

Authors: Jérôme Bolte, Edouard Pauwels

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical A mathematical model for automatic differentiation in machine learning, we provide a simple class of functions, a nonsmooth calculus, and show how they apply to stochastic approximation methods. Theorem 1 (Algorithmic differentiation does not induce an operator on functions), Theorem 2 (Algorithmic differentiation outputs a selection gradient), Theorem 3, Theorem 4 (Convergence and insignificance of artefacts)
Researcher Affiliation Academia J erˆome Bolte Toulouse School of Economics Univ. Toulouse Toulouse, France Edouard Pauwels IRIT, CNRS Univ. Toulouse Toulouse, France
Pseudocode Yes Algorithm 1: Program evaluation, Algorithm 2: Algorithmic differentiation computes selection gradients
Open Source Code No The paper does not provide any explicit statements or links indicating the availability of open-source code for the methodology described.
Open Datasets No The paper is theoretical and focuses on mathematical models and proofs. It does not describe training on a specific dataset or provide access information for any dataset used in empirical studies.
Dataset Splits No The paper is theoretical and does not describe empirical experiments, thus no dataset validation split information is provided.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory amounts) used for running its experiments.
Software Dependencies No The paper mentions 'Tensor Flow' and 'Py Torch' as general implementations but does not provide specific version numbers for software dependencies relevant to replicating its work.
Experiment Setup No The paper is theoretical and does not describe specific experimental setup details such as hyperparameter values or training configurations.