A Framework for Multistream Regression With Direct Density Ratio Estimation

Authors: Ahsanul Haque, Hemeng Tao, Swarup Chandra, Jie Liu, Latifur Khan

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

Reproducibility Variable Result LLM Response
Research Type Experimental We analyze the theoretical properties of the proposed approach and empirically evaluate it on both real-world and synthetic data sets.
Researcher Affiliation Academia Ahsanul Haque, Hemeng Tao, Swarup Chandra, Jie Liu, Latifur Khan Department of Computer Science University of Texas at Dallas, Richardson TX {ahsanul.haque, hxt160430, swarup.chandra, jxl164830, lkhan}@utdallas.edu
Pseudocode Yes Algorithm 1 Density Ratio Estimation; Algorithm 2 Drift Detection.
Open Source Code No The paper does not provide any statement or link to open-source code for the methodology described.
Open Datasets Yes We use 3 real-world and 3 synthetic data sets to evaluate the proposed framework. Table 1 lists them with corresponding properties. Real-World Data Sets The task in Power Consumption (Lichman 2013) data set is to predict the total power to be consumed by households in 2006 from readings such as reactive and active power, voltage, and intensity. In CASP ((Lichman 2013)) data set, the task is to predict the size of residue, given physiochemical properties of protein tertiary structure. Finally, the task in Airline Delay (Data Expo 2009) data set is to predict the arrival time delay of flights using features such as scheduled departure and arrival time, departure delay, and distance.
Dataset Splits No The paper describes using a 'small set of data instances from both S and T' for initial training and continuous evaluation over streams, but does not specify explicit train/validation/test splits, percentages, or cross-validation details for reproducibility in the traditional sense.
Hardware Specification No The paper does not provide specific hardware details (e.g., CPU/GPU models, memory) used for running its experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers (e.g., Python 3.8, PyTorch 1.9).
Experiment Setup Yes We use N = 300 as our default setting in the experiments. Also, λ = 0.01 and η = 1 following (Kawahara and Sugiyama 2012).