Details
Original language | English |
---|---|
Title of host publication | Advances in Knowledge Discovery and Data Mining |
Subtitle of host publication | 27th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2023, Osaka, Japan, May 25–28, 2023, Proceedings, Part IV |
Editors | Hisashi Kashima, Tsuyoshi Ide, Wen-Chih Peng |
Publisher | Springer Science and Business Media Deutschland GmbH |
Pages | 70-82 |
Number of pages | 13 |
ISBN (electronic) | 978-3-031-33383-5 |
ISBN (print) | 9783031333828 |
Publication status | Published - 26 May 2023 |
Event | 27th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2023 - Osaka, Japan Duration: 25 May 2023 → 28 May 2023 |
Publication series
Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
---|---|
Volume | 13938 LNCS |
ISSN (Print) | 0302-9743 |
ISSN (electronic) | 1611-3349 |
Abstract
Accurate spatio-temporal prediction is essential for capturing city dynamics and planning mobility services. State-of-the-art deep spatio-temporal predictive models depend on rich and representative training data for target regions and tasks. However, the availability of such data is typically limited. Furthermore, existing predictive models fail to utilize cross-correlations across tasks and cities. In this paper, we propose MetaCitta, a novel deep meta-learning approach that addresses the critical challenges of data scarcity and model generalization. MetaCitta adopts the data from different cities and tasks in a generalizable spatio-temporal deep neural network. We propose a novel meta-learning algorithm that minimizes the discrepancy between spatio-temporal representations across tasks and cities. Our experiments with real-world data demonstrate that the proposed MetaCitta approach outperforms state-of-the-art prediction methods for zero-shot learning and pre-training plus fine-tuning. Furthermore, MetaCitta is computationally more efficient than the existing meta-learning approaches.
Keywords
- Meta-Learning, Pre-training, Spatio-Temporal Prediction
ASJC Scopus subject areas
- Mathematics(all)
- Theoretical Computer Science
- Computer Science(all)
- General Computer Science
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
Advances in Knowledge Discovery and Data Mining: 27th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2023, Osaka, Japan, May 25–28, 2023, Proceedings, Part IV. ed. / Hisashi Kashima; Tsuyoshi Ide; Wen-Chih Peng. Springer Science and Business Media Deutschland GmbH, 2023. p. 70-82 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13938 LNCS).
Research output: Chapter in book/report/conference proceeding › Conference contribution › Research › peer review
}
TY - GEN
T1 - MetaCitta
T2 - 27th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2023
AU - Sao, Ashutosh
AU - Gottschalk, Simon
AU - Tempelmeier, Nicolas
AU - Demidova, Elena
N1 - Funding Information: This work was partially funded by the DFG, German Research Foundation (“WorldKG”, 424985896), the Federal Ministry for Economic Affairs and Climate Action (BMWK), Germany (“d-E-mand”, 01ME19009B), and DAAD, Germany (“KOALA”, 57600865).
PY - 2023/5/26
Y1 - 2023/5/26
N2 - Accurate spatio-temporal prediction is essential for capturing city dynamics and planning mobility services. State-of-the-art deep spatio-temporal predictive models depend on rich and representative training data for target regions and tasks. However, the availability of such data is typically limited. Furthermore, existing predictive models fail to utilize cross-correlations across tasks and cities. In this paper, we propose MetaCitta, a novel deep meta-learning approach that addresses the critical challenges of data scarcity and model generalization. MetaCitta adopts the data from different cities and tasks in a generalizable spatio-temporal deep neural network. We propose a novel meta-learning algorithm that minimizes the discrepancy between spatio-temporal representations across tasks and cities. Our experiments with real-world data demonstrate that the proposed MetaCitta approach outperforms state-of-the-art prediction methods for zero-shot learning and pre-training plus fine-tuning. Furthermore, MetaCitta is computationally more efficient than the existing meta-learning approaches.
AB - Accurate spatio-temporal prediction is essential for capturing city dynamics and planning mobility services. State-of-the-art deep spatio-temporal predictive models depend on rich and representative training data for target regions and tasks. However, the availability of such data is typically limited. Furthermore, existing predictive models fail to utilize cross-correlations across tasks and cities. In this paper, we propose MetaCitta, a novel deep meta-learning approach that addresses the critical challenges of data scarcity and model generalization. MetaCitta adopts the data from different cities and tasks in a generalizable spatio-temporal deep neural network. We propose a novel meta-learning algorithm that minimizes the discrepancy between spatio-temporal representations across tasks and cities. Our experiments with real-world data demonstrate that the proposed MetaCitta approach outperforms state-of-the-art prediction methods for zero-shot learning and pre-training plus fine-tuning. Furthermore, MetaCitta is computationally more efficient than the existing meta-learning approaches.
KW - Meta-Learning
KW - Pre-training
KW - Spatio-Temporal Prediction
UR - http://www.scopus.com/inward/record.url?scp=85173559157&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-33383-5_6
DO - 10.1007/978-3-031-33383-5_6
M3 - Conference contribution
AN - SCOPUS:85173559157
SN - 9783031333828
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 70
EP - 82
BT - Advances in Knowledge Discovery and Data Mining
A2 - Kashima, Hisashi
A2 - Ide, Tsuyoshi
A2 - Peng, Wen-Chih
PB - Springer Science and Business Media Deutschland GmbH
Y2 - 25 May 2023 through 28 May 2023
ER -