Piecewise Strong Convexity of Neural Networks

Authors: Tristan Milne

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We also include an experimental section where we validate our theoretical work and show that the regularized loss function is almost always piecewise strongly convex when restricted to stochastic gradient descent trajectories for three standard image classification problems.
Researcher Affiliation Academia Tristan Milne Department of Mathematics University of Toronto Toronto, Ontario, Canada tmilne@math.toronto.edu
Pseudocode No The paper does not contain any pseudocode or algorithm blocks.
Open Source Code No The paper does not provide any explicit statements about open-source code availability or links to repositories for the described methodology.
Open Datasets Yes For the MNIST, CIFAR10, and CIFAR100 datasets, we produce two plots generated by training neural networks using stochastic gradient descent (SGD).
Dataset Splits No The paper mentions using standard datasets like MNIST, CIFAR10, and CIFAR100 but does not explicitly provide the specific training, validation, or test split percentages or sample counts used for reproducibility.
Hardware Specification No The paper mentions the use of 'GPUs' in the acknowledgements, but does not provide specific hardware details such as GPU models, CPU types, or memory specifications used for the experiments.
Software Dependencies No The paper states that 'all experiments were implemented in Py Torch' but does not specify the version number of PyTorch or any other software dependencies, which is required for reproducibility.
Experiment Setup Yes Our experimental set-up for each data set is summarized in Table 1, and all experiments were implemented in Py Torch, with a mini-batch size of 128. Table 1 includes specific values for Epochs, Batch Norm. (Yes/No), λ (weight decay), Trials, and Learning Rate for each dataset (MNIST, CIFAR10, CIFAR100).