MetaStore: A Task-adaptative Meta-learning Model for Optimal Store Placement with Multi-city Knowledge Transfer

Abstract

Optimal store placement aims to identify the optimal location for a new brick-and-mortar store that can maximize its sale by analyzing and mining users’ preferences from large-scale urban data. In recent years, the expansion of chain enterprises in new cities brings some challenges because of two aspects: (1) data scarcity in new cities, so most existing models tend to not work (i.e., overfitting), because the superior performance of these works is conditioned on large-scale training samples; (2) data distribution discrepancy among different cities, so knowledge learned from other cities cannot be utilized directly in new cities. In this article, we propose a task-adaptative model-agnostic meta-learning framework, namely, MetaStore, to tackle these two challenges and improve the prediction performance in new cities with insufficient data for optimal store placement, by transferring prior knowledge learned from multiple data-rich cities. Specifically, we develop a task-adaptative meta-learning algorithm to learn city-specific prior initializations from multiple cities, which is capable of handling the multimodal data distribution and accelerating the adaptation in new cities compared to other methods. In addition, we design an effective learning strategy for MetaStore to promote faster convergence and optimization by sampling high-quality data for each training batch in view of noisy data in practical applications. The extensive experimental results demonstrate that our proposed method leads to state-of-the-art performance compared with various baselines.

Research on fault diagnosis model driven by artificial intelligence from domain adaptation to domain generalization

“…Therefore, fault diagnosis of key components is particularly important. According to the different health conditions of the task in the source domain, the training model [1][2][3] and the knowledge transfer learning model [4][5][6] can further improve the generalization ability of the model, break the synchronization assumption, and achieve the ability of deep transfer learning. Among them, domain adaptation includes knowledge mapping and multi-source domain adaptation; Domain generalization includes domain generalization of cross-domain confrontation training and depth domain generalization of class boundary feature detection.…”

MetaCitta: Deep Meta-Learning for Spatio-Temporal Prediction Across Cities and Tasks

“…-MetaStore [10] is a MAML-based approach that learns to generate the city-specific parameters based on the city's encoding during training. -MetaST [13] is also based on MAML and learns a pattern-based spatiotemporal memory from source cities.…”

“…Meta-learning is typically used to transfer knowledge from multiple cities (e.g., MetaStore [10] and MetaST [13]). However, existing approaches only learn from a specific task (e.g., bike demand prediction) and do not exploit the correlations between the tasks (e.g., taxi and bike demand prediction) and geographic regions.…”

“…In this paper, we propose MetaCitta, a novel deep meta-learning approach for spatio-temporal predictions across cities and tasks. In contrast to other approaches [10,13], MetaCitta adopts the knowledge not only from several cities but also different tasks. As mentioned above, different tasks in the same city can exhibit spatial correlations, and a task across cities can exhibit task-specific correlations.…”

A Cross-City Federated Transfer Learning Framework: A Case Study on Urban Region Profiling

“…• MetaStore [5]: A state-of-the-art task-adaptative modelagnostic meta-learning framework that also learns available knowledge from multiple source cities and transfers it to data-insufficiency target city to improve prediction performance on urban computing tasks. This method learns a set of meta-learned initializations from a variety of source cities, which is capable of tackling complex data distributions and accelerating the adaptation in target city.…”