Discriminative Robust Transformation Learning

Authors: Jiaji Huang, Qiang Qiu, Guillermo Sapiro, Robert Calderbank

NeurIPS 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results obtained on benchmark datasets, such as labeled faces in the wild, demonstrate the value of being able to balance discrimination and robustness.
Researcher Affiliation Academia Department of Electrical Engineering, Duke University Durham, NC 27708 {jiaji.huang,qiang.qiu,guillermo.sapiro,robert.calderbank}@duke.edu
Pseudocode Yes Algorithm 1 Gradient descent solver for Euc-DRT
Open Source Code No The paper does not provide an explicit statement or link for the open-source code of their methodology.
Open Datasets Yes Experimental results obtained on benchmark datasets, such as labeled faces in the wild, demonstrate the value of being able to balance discrimination and robustness. ... We apply a more sophisticated convolutional neural network to the MNIST dataset. ... We train a deep network on the WDRef dataset
Dataset Splits No The paper mentions training and testing sets, but does not explicitly provide details about a validation set split (percentages or counts).
Hardware Specification No The paper does not provide specific hardware details (e.g., CPU, GPU models, memory) used for running the experiments.
Software Dependencies No The paper mentions using a "D-layer neural network" and "Le Net" structure but does not provide specific version numbers for any software dependencies or libraries.
Experiment Setup Yes We use Gaussian random variables to initialize α, then, on the randomly transformed data, we set t(1) (t( 1)) to be the average intra-class (inter-class) pairwise distance. ...We calculate the diameter γ of the local regions NB indirectly, using the κ-nearest neighbors of each training sample to define a local neighborhood. ...We consider λ = 1, 0.5, 0.25...We use κ = 7 nearest neighbors to define local regions in Euc-DRT. ...Table 3: Implementation details of the neural network for MNIST classification. ...Weight decay (conventional Frobenius norm regularization) is employed in both DF and DML...