Learning-Augmented Data Stream Algorithms

Authors: Tanqiu Jiang, Yi Li, Honghao Lin, Yisong Ruan, David P. Woodruff

ICLR 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We empirically validate our results, demonstrating also our improvements in practice. We conduct experiments for the distinct elements and the Fp moment (p > 2) problems, on both real-world and synthetic data, which demonstrate significant practical benefits.
Researcher Affiliation Academia Tanqiu Jiang Department of Electrical and Computer Engineering Lehigh University Bethlehem, PA 18015, USA taj320@lehigh.edu, Yi Li School of Physical and Mathematical Sciences Nanyang Technological University Singapore 637371 yili@ntu.edu.sg, Honghao Lin Zhiyuan College Shanghai Jiao Tong University Shanghai, China 200240 honghao lin@sjtu.edu.cn, Yisong Ruan Department of Software engineering Xiamen University Xiamen, Fujian, China 361000 24320152202802@stu.edu.xmu.cn, David P. Woodruff Computer Science Department Carnegie Mellon University Pittsburgh, PA 15213, USA dwoodruf@cs.cmu.edu
Pseudocode No The paper describes algorithms such as ROUGHL0ESTIMATOR and EXACTCOUNT in textual form but does not provide pseudocode or formally labeled algorithm blocks.
Open Source Code No The paper does not contain any explicit statement or link indicating that source code for the described methodology is publicly available.
Open Datasets Yes The traffic data is collected at a backbone link of a Tier1 ISP between Chicago and Seattle in 2016 (CAIDA). http://www.caida.org/data/monitors/ passive-equinix-chicago.xml.
Dataset Splits Yes They use the first 7 minutes for training, the following minute for validation, and estimate the packet counts in subsequent minutes. [...] They use the first 5 days for training, the following day for validation, and estimate the number of times different search queries appear in subsequent days.
Hardware Specification No The paper does not specify the exact hardware components (e.g., CPU/GPU models, memory) used for running the experiments.
Software Dependencies No The paper describes the use of algorithms and models (e.g., RNNs, LSTM) but does not list specific software dependencies with version numbers.
Experiment Setup Yes For ROUGHL0ESTIMATOR, we set c = 10 and η = 1/4. We use the heavy hitter oracle to predict whether the coordinate will be larger than 210. We randomly select a prime from [11, 31] for the hash buckets. [...] We plot the results for different values of k = 10, 20, 30.