Learning to Pivot with Adversarial Networks

Authors: Gilles Louppe, Michael Kagan, Kyle Cranmer

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

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we empirically demonstrate the effectiveness of the approach with a toy example and examples from particle physics.
Researcher Affiliation Academia Gilles Louppe New York University g.louppe@nyu.edu Michael Kagan SLAC National Accelerator Laboratory makagan@slac.stanford.edu Kyle Cranmer New York University kyle.cranmer@nyu.edu
Pseudocode Yes Algorithm 1 Adversarial training of a classifier f against an adversary r.
Open Source Code Yes The source code to reproduce the experiments is available online 1. 1https://github.com/glouppe/paper-learning-to-pivot
Open Datasets Yes We reuse the datasets used in (Baldi et al., 2016a).
Dataset Splits No The paper mentions 'a subset of 150000 samples for training while AMS is evaluated on an independent test set of 5000000 samples' but does not specify a separate validation set or detailed splits (percentages, counts) for the training data or for the toy example.
Hardware Specification No The paper does not provide any specific hardware details such as GPU or CPU models, memory specifications, or types of computing resources used for experiments.
Software Dependencies No The paper does not list specific software dependencies with version numbers, such as Python versions, deep learning frameworks (e.g., PyTorch, TensorFlow), or other relevant libraries.
Experiment Setup Yes The network architecture comprises 2 dense hidden layers of 20 nodes respectively with tanh and Re LU activations, followed by a dense output layer with a single node with a sigmoid activation. ... adversarial training was performed for 200 iterations, mini-batches of size M = 128, K = 500 and λ = 50. ... The architecture of f comprises 3 hidden layers of 64 nodes respectively with tanh, Re LU and Re LU activations, and is terminated by a single final output node with a sigmoid activation. The architecture of r is the same, but uses only Re LU activations in its hidden nodes.