Multi-Resolution Weak Supervision for Sequential Data

Authors: Paroma Varma, Frederic Sala, Shiori Sagawa, Jason Fries, Daniel Fu, Saelig Khattar, Ashwini Ramamoorthy, Ke Xiao, Kayvon Fatahalian, James Priest, Christopher Ré

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We experimentally validate our framework on five real-world sequential classification tasks over modalities like medical video, gait sensor data, and industry-scale video data.
Researcher Affiliation Academia {fredsala, paroma, jfries, danfu, sagawas, saelig, ashwinir, kayvonf, jpriest, chrismre}@stanford.edu, kexiao@cs.umass.edu
Pseudocode Yes Algorithm 1: Accuracy Parameter Estimation
Open Source Code No The paper does not explicitly state that its source code is open or provide a link to a repository for its methodology.
Open Datasets Yes The Bicuspid Aortic Valve (BAV) [6] task is to classify a congenital heart defect over MRI videos from a population-scale dataset [23]... Basketball operatesoverasubsetof Activity Net[3].
Dataset Splits Yes All datasets include a small hand-labeled development set (< 10% of the unlabeled data) used to tune supervision sources and end model hyperparameters.
Hardware Specification No The paper mentions 'modern frameworks like Py Torch' but does not specify any hardware details (e.g., GPU/CPU models, memory) used for running experiments.
Software Dependencies No The paper mentions 'Py Torch' but does not provide a specific version number or other software dependencies with version numbers.
Experiment Setup Yes Results are reported on test set as the mean S.D. of F1 scores across 5 random weight initializations.