Statistical mechanics of low-rank tensor decomposition

Authors: Jonathan Kadmon, Surya Ganguli

NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our theory reveals the existence of phase transitions between easy, hard and impossible inference regimes, and displays an excellent match with simulations. Finally, we compare our AMP algorithm to the most commonly used algorithm, alternating least squares (ALS), and demonstrate that AMP significantly outperforms ALS in the presence of noise.
Researcher Affiliation Collaboration Jonathan Kadmon Department of Applied Physics, Stanford University kadmonj@stanford.edu Surya Ganguli Department of Applied Physics, Stanford University and Google Brain, Mountain View, CA sganguli@stanford.edu
Pseudocode No The paper presents iterative update equations (11)-(14) but does not format them within a distinct pseudocode block or algorithm environment.
Open Source Code Yes Code to reproduce all simulations presented in this paper is available at https://github.com/ganguli-lab/tensor AMP.
Open Datasets No The paper mentions generating "order-3 tensors generated randomly according to (2)" for simulations, implying synthetic data without providing access information or formal citations for a publicly available dataset used for training, validation, or testing.
Dataset Splits No The paper describes numerical simulations but does not specify train/validation/test dataset splits, percentages, or sample counts for experimental reproduction.
Hardware Specification No No specific hardware details (e.g., GPU/CPU models, memory, cloud instance types) used for running experiments are provided in the paper.
Software Dependencies No The paper does not specify any software dependencies with version numbers (e.g., specific libraries, frameworks, or programming language versions) that are critical for reproducing the experiments.
Experiment Setup Yes Figure 2.A caption: "[σα = 1, µ1 = µ2 = 0.1 (blue), µ3 = 0.3 (orange), N = 500, nα = 1]". Figure 3 caption: "[p = 3, r = 1, σα = 1, µα = 0.2, nα = {1, 8/8 }, N = 500 ]"