SUBSET CONDITIONING USING VARIATIONAL AUTOENCODER WITH A LEARNABLE TENSOR TRAIN INDUCED PRIOR

Invention Publication

US20240152763A1 SUBSET CONDITIONING USING VARIATIONAL AUTOENCODER WITH A LEARNABLE TENSOR TRAIN INDUCED PRIOR 审中-公开

Please log in to see more content

Patent Title: SUBSET CONDITIONING USING VARIATIONAL AUTOENCODER WITH A LEARNABLE TENSOR TRAIN INDUCED PRIOR
Application No.: US18538870

Application Date: 2023-12-13
Publication No.: US20240152763A1

Publication Date: 2024-05-09
Inventor: Aleksandr Aliper , Aleksandrs Zavoronkovs , Alexander Zhebrak , Daniil Polykovskiy , Maksim Kuznetsov , Yan Ivanenkov , Mark Veselov , Vladimir Aladinskiy , Evgeny Putin , Yuriy Volkov , Arip Asadulaev
Applicant: INSILICO MEDICINE IP LIMITED
Applicant Address: HK Hong Kong
Assignee: INSILICO MEDICINE IP LIMITED
Current Assignee: INSILICO MEDICINE IP LIMITED
Current Assignee Address: HK Hong Kong
Main IPC: G06N3/092
IPC: G06N3/092 ; G06N3/0455 ; G06N3/0475 ; G06N3/084

SUBSET CONDITIONING USING VARIATIONAL AUTOENCODER WITH A LEARNABLE TENSOR TRAIN INDUCED PRIOR

Abstract:

The proposed model is a Variational Autoencoder having a learnable prior that is parametrized with a Tensor Train (VAE-TTLP). The VAE-TTLP can be used to generate new objects, such as molecules, that have specific properties and that can have specific biological activity (when a molecule). The VAE-TTLP can be trained in a way with the Tensor Train so that the provided data may omit one or more properties of the object, and still result in an object with a desired property.

Information query

Global Dossier Espacenet