An Accelerated Proximal Coordinate Gradient Method

Authors: Qihang Lin, Zhaosong Lu, Lin Xiao

NeurIPS 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We present numerical experiments to illustrate the advantage of our method.
Researcher Affiliation Collaboration Qihang Lin University of Iowa Iowa City, IA, USA qihang-lin@uiowa.edu Zhaosong Lu Simon Fraser University Burnaby, BC, Canada zhaosong@sfu.ca Lin Xiao Microsoft Research Redmond, WA, USA lin.xiao@microsoft.com
Pseudocode Yes Algorithm 1: the APCG method... Algorithm 2: APCG with µ = 0... Algorithm 3: APCG with γ0 = µ > 0... Algorithm 4: Efficient implementation of APCG with γ0 = µ > 0... Algorithm 5: APCG for solving dual ERM
Open Source Code No The paper does not contain any statements about making the source code available, nor does it provide a link to a code repository for the methodology described.
Open Datasets Yes The dataset used in our experiments are summarized in Table 1. ... (available from the LIBSVM web page: http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets).
Dataset Splits No The paper mentions the datasets used (rcv1, covertype, news20) but does not specify the training, validation, or test splits. It does not provide percentages, sample counts, or refer to standard predefined splits with a citation to allow reproduction of the data partitioning.
Hardware Specification No The paper does not explicitly describe the hardware used for its experiments. It does not mention specific GPU models, CPU models, or detailed system specifications (e.g., memory, cloud instance types).
Software Dependencies No The paper mentions comparing with "SDCA" and "accelerated full gradient method (AFG)" but does not provide specific version numbers for any software, libraries, or programming languages used in the implementation or experiments. Thus, it does not provide reproducible ancillary software details.
Experiment Setup Yes In our experiments, we solve ERM problems with smoothed hinge loss for binary classification. ... The conjugate function of φ is φ (b) = b + γ/2 b2 if b ∈ [−1, 0] and otherwise. ... with λ varying form 10^−5 to 10^−8.