Title:
AUTOMATED MACHINE LEARNING SYSTEM AND METHOD FOR STREAMING DATA ANALYSIS FOR USE IN CONCEPTUAL TRANSFER
Document Type and Number:
WIPO Patent Application WO/2020/155594
Kind Code:
A1
Abstract:
Disclosed are an automated machine learning system and method for streaming data analysis for use in conceptual transfer. The system comprises: an acquisition module, used for acquiring each batch of serialized streaming datasets; a generating module, used for automated feature processing, self-adaptive conceptual migration, and hyperparameter optimization with respect to each batch of serialized streaming datasets; a feedback module, used for inputting a preset batch of serialized datasets into a prediction model to generate a feedback result; and an online adjustment module, used for online adjustment of the prediction model on the basis of the feedback result to implement automated machine learning. The system automatedly selects feature processing techniques with respect to different application scenarios, self-adapts to conceptual transfers, has great performance in real scenarios, automatedly compresses a search space for automated learning by means of a resource control technique, and ensures that the scale of model computing matches the scale and features of data.
More Like This:
Inventors:
ZHU WENWU (CN)
XIONG ZHENG (CN)
JIANG JIYAN (CN)
ZHANG WENPENG (CN)
XIONG ZHENG (CN)
JIANG JIYAN (CN)
ZHANG WENPENG (CN)
Application Number:
PCT/CN2019/098828
Publication Date:
August 06, 2020
Filing Date:
August 01, 2019
Export Citation:
Assignee:
UNIV TSINGHUA (CN)
International Classes:
G06N20/00
Foreign References:
| CN109858631A | 2019-06-07 | |||
| CN106779087A | 2017-05-31 | |||
| CN101782976A | 2010-07-21 | |||
| US20080059395A1 | 2008-03-06 |
Attorney, Agent or Firm:
TSINGYIHUA INTELLECTUAL PROPERTY LLC (CN)
Download PDF: