Meta-Learning Framework with Applications to Zero-Shot Time-Series Forecasting
Authors: Boris N. Oreshkin, Dmitri Carpov, Nicolas Chapados, Yoshua Bengio9242-9250
AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our empirical results on a wide range of data emphasize the importance of the identified meta-learning mechanisms for successful zero-shot univariate forecasting, suggesting that it is viable to train a neural network on a source TS dataset and deploy it on a different target TS dataset without retraining, resulting in performance that is at least as good as that of state-of-practice univariate forecasting models. |
| Researcher Affiliation | Collaboration | Boris N. Oreshkin1, Dmitri Carpov1, Nicolas Chapados1, Yoshua Bengio2 1Element AI, 2Mila |
| Pseudocode | No | The paper provides equations and describes the N-BEATS architecture, but it does not present a formal pseudocode block or algorithm box. |
| Open Source Code | No | The paper states it makes 'dataset loaders and evaluation code public' for a new task/dataset, but it does not explicitly state that the source code for the main methodology (N-BEATS) itself is publicly available or provide a link to it. |
| Open Datasets | Yes | Base datasets. M4 (M4 Team 2018), contains 100k TS representing demographic, finance, industry, macro and micro indicators. M3 (Makridakis and Hibon 2000) contains 3003 TS from domains and sampling frequencies similar to M4. FRED is a dataset introduced in this paper containing 290k US and international economic TS from 89 sources, a subset of the data published by the Federal Reserve Bank of St. Louis (Federal Reserve 2019). TOURISM (Athanasopoulos et al. 2011) includes monthly, quarterly and yearly series of indicators related to tourism activities. ELECTRICITY (Dua and Graff 2017; Yu, Rao, and Dhillon 2016) represents the hourly electricity usage of 370 customers. TRAFFIC (Dua and Graff 2017; Yu, Rao, and Dhillon 2016) tracks hourly occupancy rates of 963 lanes in the Bay Area freeways. |
| Dataset Splits | Yes | The forecasted TS is split into two non-overlapping pieces: the history, and the test. The history is used as model input and the test is used to compute the forecast error metric. We use the history and the test splits for the base datasets consistent with their original publication, unless explicitly stated otherwise. |
| Hardware Specification | No | The paper describes the models, data, and experimental setup, but does not specify any particular hardware (e.g., GPU/CPU models, memory) used for running the experiments. |
| Software Dependencies | No | Deep AR (Salinas et al. 2019) is trained using Gluon TS implementation from its authors (Alexandrov et al. 2019). N-BEATS is trained following the original training setup of Oreshkin et al. (2020). While Gluon TS is mentioned, no specific version numbers are provided for it or any other software dependencies. |
| Experiment Setup | Yes | Deep AR (Salinas et al. 2019) is trained using Gluon TS implementation from its authors (Alexandrov et al. 2019). N-BEATS is trained following the original training setup of Oreshkin et al. (2020). Both N-BEATS and Deep AR are trained with scaling/descaling the architecture input/output by dividing/multiplying all input/output values by the max value of the input window computed per target time-series. [...] Additional training setup details are provided in Appendix D. |