0.25
0.5
0.75
1.25
1.5
1.75
2
Machine learning for environmental and life sciences
Published on 2019-03-27265 Views
Increasingly often, we need to learn predictive models from big or complex data, which may comprise many examples and many input/output dimensions. When more than one target variable has to be predict
Related categories
Presentation
Machine Learning for Environmental & Life Sciences00:00
What is artificial intelligence?00:37
Machine Learning01:45
Data Science02:50
The most popular data science topic04:53
Machine Learning of Interpretable Models06:33
The basic Machine Learning task: Predictive modeling06:53
An example task of Predictive Modelling: Medical diagnosis07:20
Example task: Descriptive vars.; Biomarkers for Alzheimer’s07:48
Example: Decision tree for diagnosis08:14
Another example of single-target predictive modeling (classification) - 108:28
Another example of single-target predictive modeling (classification) - 209:11
What is a decision tree?09:39
Making a Prediction with a Decision Tree10:22
Machine Learning of Decision Trees11:45
Top-Down Induction of Decision Trees15:02
TDIDT Illustrated15:06
Mining Big and Complex Data:Dimensions of Complexity17:01
Mining Big and Complex Data17:45
Big Data: Variety - Structured Input - 119:07
Big Data: Variety - Structured Input - 219:46
Predictive modeling: Structured output20:16
Big Data: Volume & Velocity20:37
Data streams: Regression21:11
Semi-supervised learning: Classification and regression21:32
Data in context: Spatio-temporal, network22:59
The Different Tasks of Multi-Target Prediction23:54
Weather prediction24:11
Multi-target prediction25:04
Example MTR task: Target vars.; Clinical scores for Alzheimer’s25:31
Example MTR model26:23
Multi-Target Classification & Multi-Label Classification27:03
Multi-Label Classification Example27:40
Hierarchical multi-label classification28:24
Hierarchical multi-label classif.28:36
Hierarchical multi-label classification: Another example30:00
The hierarchy can be a tree or a DAG30:21
Hierarchical structure on target space for the ADNI dataset30:51
Mining Big and Complex Data:Combining Complexities31:37
Data streams: (MT) Regression32:08
Network +SOP: HMC32:16
Even more complicated tasks33:38
Predictive Clustering for Multi-Target Prediction33:53
Clustering34:29
Example predictive clustering tree35:17
Top-down induction of PCTs36:47
Predictive clustering38:51
Selecting the best test in a PCT39:42
Multi-target regression39:52
Multi-target classification40:02
Ensembles of PCTs40:18
SSL+SOP: Incomplete Annotations41:20
Learning PCTs41:36
Relating the Environment and the Biota41:50
Environment <-> Biota42:46
Habitat modeling43:04
Predicting species composition43:12
Predicting community structure43:19
Slovenian rivers43:28
Danish farms: Soil Microarthropods43:58
Victoria, AustraliaVegetation44:03
Slovenian rivers: Species comp.44:30
Slovenian rivers: Habitat models44:56
Slovenian rivers: Community struc.45:10
Community structure: Overall results45:26
Predicting Gene Functions48:50
Predicting gene functions49:01
Predicting Gene Functions in Bacterial Genomes (RBI+JSI)49:34
GFP Pipeline50:37
Different Features Sets for GFP51:27
Gene Function Prediction: Predictive Performance52:03
Metagenome Phyletic Profiles52:50
Multi-Target Prediction for Virtual Compound Screening54:09
Virtual compound screening54:25
Host-targeted Drugs for MTB (Tuberculosis) and STM (Salmonella)56:20
MTB&STM: Host-targeted Drugs57:06
MTB&STM: Host-targeted Drugs The Data Analysis Workflow57:43
MTB&STM: Host-targeted DrugsResults58:40
Analyzing data from High-contents Screens58:58
HTS: Modulating fibroblast to myofibroblast transition59:33
Hits in the HTS screen59:43
Reducing fibrosis in myocardial infarction59:53
Testing the predictions01:00:07
Spring school in Bled in May01:00:51
Conclusions01:01:37