Multi-task learning in the analysis of phenotypic data thumbnail
Pause
Mute
Subtitles not available
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Multi-task learning in the analysis of phenotypic data

Published on Jun 28, 201984 Views

Multi-task learning is an efficient approach of machine learning which combines data from several related tasks, improving accuracy compared to solving each task separately. With the special requireme

Related categories

Chapter list

Multi-task learning in the analysis of phenotypic data00:00
Overview00:08
Multi-task learning00:57
Motivation01:03
Recommender systems - 104:20
Recommender systems - 205:45
Matrix Factorization - 106:36
Matrix Factorization - 206:41
Matrix Factorization - 308:22
Results on Netflix challenge09:15
Matrix Factorization - 410:24
Cold start problem10:29
Matrix factorization with side information - 110:38
Matrix factorization with side information - 211:30
Beyond 2-way relations12:17
Matrix Factorization - 513:38
Probabilistic Model13:56
Matrix Factorization - 616:45
Deep Learning17:08
Embedding and one-hot encoding18:34
Deep factorization model21:26
Application: Chemogenomics22:10
Chemogenomics22:16
Series effect22:57
Evaluation results25:31
Application: Binding mode27:53
Affinity and potency27:57
Competitive vs. non-competitive inhibition28:34
Identification of binding mode30:30
The setup31:19
Latent dimensions for IC50 and Ki32:17
Predicting interaction of pairs33:08
Predicting dominant behaviour33:49
Possible future work35:29
Cell Chemical Biology36:26
Classical high-content imaging36:54
Predicting unrelated protein assays37:29
Assay preparation38:23
Machine Learning Approaches39:14
Experimental Results39:25
In vitro validation40:52
Combining Multiple cell lines - 142:32
Combining Multiple cell lines - 243:02
Number of well predicted targets43:22
Significant differences in recognized assays44:18
Data fusion results45:31
Discussion and future work46:30
Scientific reports48:17
Computational setup48:51
Possible questions50:27
Feature interaction analysis - 151:11
Feature interaction analysis - 252:34
Packages53:41
Thank you fo your attention!56:40