Variance-based Regularization with Convex Objectives

Authors: Hongseok Namkoong, John C. Duchi

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

Reproducibility Variable Result LLM Response
Research Type Experimental We give corroborating empirical evidence showing that in practice, the estimator indeed trades between variance and absolute performance on a training sample, improving out-of-sample (test) performance over standard empirical risk minimization for a number of classification problems.
Researcher Affiliation Academia Hongseok Namkoong Stanford University hnamk@stanford.edu John C. Duchi Stanford University jduchi@stanford.edu
Pseudocode No The paper does not contain any pseudocode or algorithm blocks.
Open Source Code Yes Code is available at https://github.com/hsnamkoong/robustopt.
Open Datasets Yes For our first experiment, we compare our robust regularization procedure to other regularizers using the HIV-1 protease cleavage dataset from the UCI ML-repository [14]. [...] The Reuters RCV1 Corpus [13] has 804,414 examples with d = 47,236 features
Dataset Splits Yes For validation, we perform 50 experiments, where in each experiment we randomly select 9/10 of the data to train the model, evaluating its performance on the held out 1/10 fraction (test). [...] We partition the data into ten equally-sized sub-samples and perform ten validation experiments, where in each experiment we use one of the ten subsets for fitting the logistic models and the remaining nine partitions as a test set to evaluate performance.
Hardware Specification No The paper does not specify the hardware used for running the experiments (e.g., CPU, GPU, or memory details).
Software Dependencies No The paper mentions using 'logistic loss' and 'elastic net regularization' but does not specify software names with version numbers for reproducibility.
Experiment Setup Yes We use the logistic loss ( ; (x, y)) = log(1 + exp( y >x)). We compare the performance of different constraint sets by taking = 2 Rd : a1 k k1 + a2 k k2 r , which is equivalent to elastic net regularization [27], while varying a1, a2, and r. We experiment with 1-constraints (a1 = 1, a2 = 0) with r 2 {50, 100, 500, 1000, 5000}, 2-constraints (a1 = 0, a2 = 1) with r 2 {5, 10, 50, 100, 500}, elastic net (a1 = 1, a2 = 10) with r 2 {102, 2 102, 103, 2 103, 104}, our robust regularizer with 2 {102, 103, 104, 5 104, 105} and our robust regularizer coupled with the 1-constraint (a1 = 1, a2 = 0) with r = 100.