Time Series Anomaly Detection with Multiresolution Ensemble Decoding

Authors: Lifeng Shen, Zhongzhong Yu, Qianli Ma, James T. Kwok9567-9575

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

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive empirical studies on real-world benchmark data sets demonstrate that the proposed RAMED model outperforms recent strong baselines on time series anomaly detection.
Researcher Affiliation Academia 1 Department of Computer Science and Enginering, Hong Kong University of Science and Technology, Hong Kong 2 School of Computer Science and Engineering, South China University of Technology, Guangzhou 3 Key Laboratory of Big Data and Intelligent Robot (South China University of Technology), Ministry of Education
Pseudocode Yes Algorithm 1 Recurrent Autoencoder with Multiresolution Ensemble Decoding (RAMED).
Open Source Code No The paper provides links to baseline implementations (RAE, RRN, Beat GAN, RAE-ensemble) but does not provide a link or explicit statement about the availability of the source code for their proposed RAMED model.
Open Datasets Yes ECG, 2D-gesture and Power-demand are from http://www.cs.ucr.edu/ eamonn/discords/, while Yahoo s S5 is from https://webscope.sandbox.yahoo.com/.
Dataset Splits Yes We use 30% of the training set for validation, and the rest for actual training. The model with the lowest reconstruction loss on the validation set is selected for evaluation. For Yahoo s S5, the available data set is split into three parts: with 40% of the samples for training, another 30% for validation, and the remaining 30% for testing.
Hardware Specification No The paper does not provide any specific details about the hardware used to run the experiments (e.g., GPU models, CPU types, or cloud computing resources).
Software Dependencies No The paper mentions the 'Adam optimizer (Kingma and Ba 2015)' but does not provide specific version numbers for Adam or any other software libraries or dependencies used in the experiments.
Experiment Setup Yes We use 3 encoders and 3 decoders. Each encoder and decoder is a single-layer LSTM with 64 units. We perform grid search on the hyperparameter β in (7) from {0.1, 0.2, . . . , 0.9}, λ in (12) from {10 4, 10 3, 10 2, 10 1, 1}, τ in (6) is set to 3 and γ in (10) is set to 0.1. The Adam optimizer (Kingma and Ba 2015) is used with an initial learning rate of 10 3.