US Patent Application 18204568. GRADIENT FLOWS IN DATASET SPACE simplified abstract
Contents
GRADIENT FLOWS IN DATASET SPACE
Organization Name
Microsoft Technology Licensing, LLC==Inventor(s)==
[[Category:David Alvarez-melis of Somerville MA (US)]]
[[Category:Nicolo Fusi of Watertown MA (US)]]
GRADIENT FLOWS IN DATASET SPACE - A simplified explanation of the abstract
This abstract first appeared for US patent application 18204568 titled 'GRADIENT FLOWS IN DATASET SPACE
Simplified Explanation
- The patent application discusses devices, systems, and methods for machine learning (ML) using datasets. - The method involves receiving a request to operate on a first dataset that contains pairs of features and labels. - Multiple datasets are available, and the method identifies a second dataset that also contains pairs of features and labels. - The method then determines the distance between the pairs of features and labels in the first and second datasets. - Based on this determined distance, the method flows the first dataset using a dataset objective to generate an optimized dataset. - The objective is to optimize the dataset by considering the distance between the feature and label pairs in the first and second datasets.
Original Abstract Submitted
Generally discussed herein are devices, systems, and methods for machine learning (ML) by flowing a dataset towards a target dataset. A method can include receiving a request to operate on a first dataset including first feature, label pairs, identifying a second dataset from multiple datasets, the second dataset including second feature, label pairs, determining a distance between the first feature, label and the second feature, label pairs, and flowing the first dataset using a dataset objective that operates based on the determined distance to generate an optimized dataset.
