Adversarial Robustness of Streaming Algorithms through Importance Sampling

Authors: Vladimir Braverman, Avinatan Hassidim, Yossi Matias, Mariano Schain, Sandeep Silwal, Samson Zhou

NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, we empirically confirm the robustness of our algorithms on various adversarial attacks and demonstrate that by contrast, some common existing algorithms are not robust.
Researcher Affiliation Collaboration Vladimir Braverman Google vbraverman@google.com Avinatan Hassidim Google avinatan@google.com Yossi Matias Google yossi@google.com Mariano Schain Google marianos@google.com Sandeep Silwal MIT silwal@mit.edu Samson Zhou Carnegie Mellon University samsonzhou@gmail.com
Pseudocode Yes Algorithm 1 Row sampling based algorithms, e.g., [CMP16, BDM+20]
Open Source Code No The paper mentions using and comparing against existing implementations and their 'own row sampling based implementation' but does not provide any link or explicit statement about making their source code publicly available.
Open Datasets No The paper describes generating its own data for adversarial settings (e.g., 'randomly sampled from a two dimensional standard normal distribution'), but it does not provide access information (link, DOI, citation) for a publicly available or open dataset.
Dataset Splits No The paper describes the data generation for its adversarial experiments but does not specify training, validation, or test split percentages, sample counts, or reference predefined splits with citations for reproducibility.
Hardware Specification No The paper does not provide any specific details about the hardware (e.g., GPU/CPU models, memory) used to run the experiments.
Software Dependencies No The paper mentions using specific software like 'Spark' [ZXW+16b], 'Stream KM++' [AMR+12], and 'River' [MHM+20] but does not provide specific version numbers for these or any other software dependencies, which are necessary for reproducibility.
Experiment Setup No The paper describes the adversarial data generation process for its experiments but does not provide specific hyperparameters (e.g., learning rate, batch size, epochs) or detailed training configurations for the algorithms used or compared.