Topology and Geometry of Half-Rectified Network Optimization

Authors: C. Daniel Freeman, Joan Bruna

ICLR 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Section 3 presents our path discovery algorithm and Section 4 covers the numerical experiments. For our numerical experiments, we calculated normalized geodesic lengths for a variety of regression and classification tasks.
Researcher Affiliation Academia C. Daniel Freeman Department of Physics University of California at Berkeley Berkeley, CA 94720, USA daniel.freeman@berkeley.edu Joan Bruna Courant Institute of Mathematical Sciences New York University New York, NY 10011, USA bruna@cims.nyu.edu
Pseudocode Yes Algorithm 1 Greedy Dynamic String Sampling
Open Source Code Yes For more complete architecture and implementation details, see our Git Hub page.
Open Datasets Yes Our algorithm uses dynamic programming and can be efficiently deployed to study mid-scale CNN architectures on MNIST, CIFAR-10 and RNN models on Penn Treebank next word prediction.
Dataset Splits No The paper does not explicitly provide details about training/validation dataset splits, such as percentages or sample counts, distinct from the test set.
Hardware Specification No The paper does not provide specific details about the hardware used for experiments (e.g., CPU/GPU models, memory).
Software Dependencies No The paper does not specify software dependencies with version numbers (e.g., Python, PyTorch, TensorFlow versions).
Experiment Setup Yes We studied a 1-4-4-1 fully connected multilayer perceptron style architecture with sigmoid nonlinearities and RMSProp/ADAM optimization.