Learning privately from multiparty data

Authors: Jihun Hamm, Yingjun Cao, Mikhail Belkin

ICML 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We demonstrate the performance of our method with realistic tasks of activity recognition, network intrusion detection, and malicious URL detection. and We evaluate the methods with real tasks in Sec. 6 and conclude the paper in Sec. 7.
Researcher Affiliation Academia Jihun Hamm HAMMJ@CSE.OHIO-STATE.EDU Dept. Computer Science and Engineering, Ohio State University, Columbus, OH 43210, USA and Paul Cao YIC242@ENG.UCSD.EDU Dept. Computer Science and Engineering, UC-San Diego, La Jolla, CA 92093, USA and Mikhail Belkin MBELKIN@CSE.OHIO-STATE.EDU Dept. Computer Science and Engineering, Ohio State University, Columbus, OH 43210, USA
Pseudocode Yes Algorithm 1 DP Ensemble by Majority-voted ERM and Algorithm 2 DP Ensemble by Weighted ERM
Open Source Code No The paper does not provide any statement or link indicating the availability of open-source code for the methodology described.
Open Datasets Yes We use the UCI Human Activity Recognition Dataset (Anguita et al., 2012), which is a collection of motion sensor data on a smart device by multiple subjects performing 6 activities... and For this task we use the KDD-99 dataset, which consists of examples of bad connections... and The Malicious URL Dataset (Ma et al., 2009) is a collection of examples of malicious URLs...
Dataset Splits No The paper specifies training and testing sample sizes (e.g., 'The training and testing samples are 7K and 3K, respectively.' for Activity Recognition) but does not explicitly mention a separate validation split or how model selection was performed using validation data.
Hardware Specification No The paper does not provide specific details about the hardware (e.g., GPU/CPU models, memory) used for running the experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers (e.g., library names, programming language versions, or specific solver versions) used in the experiments.
Experiment Setup Yes The only hyperparameter of the model is the regularization coefficient λ which we fixed to 10 4 after performing some preliminary experiments.