prosem2/topics_iml.csv

1	index	title	chapter	who	mail	name	note1	note2
2	1	Why do we care about IML?	3.1-3.3	Daniel	daniel.wilmes@cs.uni-dortmund.de	1	e.g. Is IML required?	{null}
3	2	How to do IML research?	3.4-3.6	Daniel	daniel.wilmes@cs.uni-dortmund.de	2	e.g. How to evaluate Interpretability	{null}
4	3	Linear Models	5.1-5.3	Jelle	jelle.huentelmann@cs.tu-dortmund.de	3	Simple Models are simple to explain	Programming task: Do a linear regression on a simple dataset!
5	4	Decision Trees	5.4	Jelle	jelle.huentelmann@cs.tu-dortmund.de	4	Programming task: Train a decision Tree on a simple dataset!	Could be combined with 5
6	5	Rule Based Methods	5.5-5.6	Jelle	jelle.huentelmann@cs.tu-dortmund.de	5	{null}	Could be combined with 4
7	6	Partial Dependence Plot	8.1	Carina	carina.newen@cs.uni-dortmund.de	6	How much does chainging a feature change the output?	Could be combined with 7
8	7	Accumulated Local Effects	8.2	Carina	carina.newen@cs.uni-dortmund.de	7	How much effect does changing a feature have on the average prediction	Could be combined with 6
9	8	Feature Interactions	8.3	Carina	carina.newen@cs.uni-dortmund.de	8	In general features are not independent	Measure the effect of interactions between them
10	9	Functional Decomposition	8.4	Daniel	daniel.wilmes@cs.uni-dortmund.de	9	Describe a function by feature interactions and their interactions	{null}
11	10	Permutation Feature Importance	8.5	Bin	bin.li@tu-dortmund.de	10	How much does a feature change, if we permute its values	{null}
12	11	Global Surrogates	8.6	Bin	bin.li@tu-dortmund.de	11	Replace a complicated model by an interpretable one	{null}
13	12	Prototypes	8.7	Bin	bin.li@tu-dortmund.de	12	Represent some model output by well fitting data instances	{null}
14	13	Individual Conditional Expectation	9.1-9.2	Chiara	chiara.balestra@cs.uni-dortmund.de	13	Show the effect one feature has on the prediction	{null}
15	14	Counterfactual Explanations	9.3-9.4	Chiara	chiara.balestra@cs.uni-dortmund.de	14	What to do to change a prediction?	{null}
16	15	Shapley Values	9.5-9.6	Chiara	chiara.balestra@cs.uni-dortmund.de	15	Use game theory to explain the output of a model	{null}
17	16	Learned Features	10.1	Benedikt	benedikt.boeing@cs.tu-dortmund.de	16	Conv. NN contain Intepretable Features	Programming task: Visualise your own classifier!
18	17	Saliency Maps	10.2	Simon	simon.kluettermann@cs.uni-dortmund.de	17	Different parts of an image have different effect/importance on the classification of an image	Programming task: Generate one Saliency Map yourself!
19	18	Concept Detection	10.3	Simon	simon.kluettermann@cs.uni-dortmund.de	18	Replace Features by Concepts	{null}
20	19	Adversarials	10.4	Benedikt	benedikt.boeing@cs.tu-dortmund.de	19	Slight changes in a neural network can change its output drastically	{null}
21	20	Influential Instances	10.5	Simon	simon.kluettermann@cs.uni-dortmund.de	20	Single examples can change the output of a NN drastically	{null}