Skip to content

Commit 3f75b15

Browse files
pat-altpat-alt
committed
set up template for paper
1 parent dea7460 commit 3f75b15

12 files changed

Lines changed: 107 additions & 214 deletions

File tree

‎_quarto.yml‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,6 @@
11
project:
22
title: "Endogenous Dynamics in Algorithmic Recourse"
3+
output-dir: build
34
filters:
45
- lua/abstract-to-meta.lua
56
- quarto

‎paper/paper.html‎

Lines changed: 0 additions & 168 deletions
This file was deleted.

‎paper/paper.pdf‎

-24.1 KB
Binary file not shown.

‎paper/paper.qmd‎

Lines changed: 54 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -1,24 +1,64 @@
11
---
2-
title: "Overlooked Dynamics in Algorithmic Recourse"
2+
title: "Endogenous Dynamics in Algorithmic Recourse"
3+
author: |
4+
\author{\IEEEauthorblockN{Patrick Altmeyer}
5+
\IEEEauthorblockA{\textit{Delft University of Technology} \\
6+
\textit{EEMCS}\\
7+
Delft, Netherlands \\
8+
p.altmeyer[at]tudelft.nl}
9+
\and
10+
\IEEEauthorblockN{Cynthia C. S. Liem}
11+
\IEEEauthorblockA{\textit{Delft University of Technology} \\
12+
\textit{EEMCS}\\
13+
Delft, Netherlands \\
14+
c.c.s.liem[at]tudelft.nl}
15+
\and
16+
\IEEEauthorblockN{3\textsuperscript{rd} Given Name Surname}
17+
\IEEEauthorblockA{\textit{dept. name of organization (of Aff.)} \\
18+
\textit{name of organization (of Aff.)}\\
19+
City, Country \\
20+
email address or ORCID}
21+
\and
22+
\IEEEauthorblockN{4\textsuperscript{th} Given Name Surname}
23+
\IEEEauthorblockA{\textit{dept. name of organization (of Aff.)} \\
24+
\textit{name of organization (of Aff.)}\\
25+
City, Country \\
26+
email address or ORCID}
27+
\and
28+
\IEEEauthorblockN{5\textsuperscript{th} Given Name Surname}
29+
\IEEEauthorblockA{\textit{dept. name of organization (of Aff.)} \\
30+
\textit{name of organization (of Aff.)}\\
31+
City, Country \\
32+
email address or ORCID}
33+
\and
34+
\IEEEauthorblockN{6\textsuperscript{th} Given Name Surname}
35+
\IEEEauthorblockA{\textit{dept. name of organization (of Aff.)} \\
36+
\textit{name of organization (of Aff.)}\\
37+
City, Country \\
38+
email address or ORCID}
39+
}
340
format:
441
pdf:
42+
documentclass: IEEEtran
43+
classoption: conference
544
include-in-header: tex/preamble.tex
6-
include-before-body: tex/before_body.tex
7-
html: default
8-
link-citations: true
45+
csl: https://raw.githubusercontent.com/citation-style-language/styles/master/ieee.csl
46+
number-sections: true
47+
license: "CC BY"
948
---
1049

11-
```{.include}
12-
sections/abstract.qmd
13-
```
50+
{{< include sections/abstract.qmd >}}
1451

15-
```{.include}
16-
sections/introduction.qmd
17-
```
52+
{{< include sections/introduction.qmd >}}
1853

19-
@schut2021generating
54+
# Acknoledgment {.unnumbered}
2055

21-
```{.include}
22-
sections/appendix.qmd
23-
```
56+
P. A. thanks ...
57+
58+
# References {.unnumbered}
59+
60+
::: {#refs}
61+
:::
62+
63+
{{< include sections/appendix.qmd >}}
2464

‎paper/sections/abstract.qmd‎

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1 +1,3 @@
1+
# Abstract
2+
13
Existing work on Counterfactual Explanations (CE) and Algorithmic Recourse (AR) has largely been limited to the static setting: given some classifier we are interested in finding close, actionable, realistic, sparse, diverse and ideally causally founded counterfactuals. The ability of CE to handle dynamics like data and model drift remains a largely unexplored research challenge at this point. Only one recent work considers the implications of exogenous domain and model shifts. This project instead focuses on endogenous dynamics, that is shifts that occur when AR is actually implemented by a proportion of individuals. Early findings suggest that the involved shifts may be large with important implications on the validity of AR and the overall characteristics of the sample population.

‎paper/sections/appendix.qmd‎

Lines changed: 3 additions & 11 deletions
Original file line numberDiff line numberDiff line change
@@ -2,31 +2,23 @@
22
format: gfm
33
---
44

5-
\pagebreak
6-
7-
## References {.appendix}
85

9-
::: {#refs}
10-
:::
116

12-
\FloatBarrier
137
\pagebreak
148

15-
## Tables {#tables .appendix}
9+
# Tables {#tables .appendix}
1610

1711
...
1812

19-
\FloatBarrier
2013
\pagebreak
2114

22-
## Figures {#figures .appendix}
15+
# Figures {#figures .appendix}
2316

2417
...
2518

26-
\FloatBarrier
2719
\pagebreak
2820

29-
## Code {#code .appendix}
21+
# Code {#code .appendix}
3022

3123
...
3224

‎paper/sections/introduction.qmd‎

Lines changed: 3 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,5 @@
1+
# Introduction {#intro}
2+
13
Recent advances in Artificial Intelligence (AI) have propelled its adoption in domains outside of Computer Science including Healthcare, Bioinformatics and Genetics. In Finance, Economics and other social sciences, applications of AI are still relatively limited. Decision-making in these fields has traditionally been guided by interpretable models that facilitate explanations. Explainability is crucial in this context, since decision-makers are typically held accountable by the public: central banks, for example, are heavily scrutinized for the policies they impose. It is therefore not surprising that practitioners and academics in these fields are reluctant to adopt AI technologies they cannot trust. Deep neural networks, for example, are generally considered as black boxes and therefore not trustworthy in a context that demands explanations.
24

35
**Counterfactual explanations** (CE) explain how inputs into a model need to change for it to produce different outputs. They are intuitive, simple and intrinsically linked to the potential outcome framework for causal inference, which social scientists are familiar with. Counterfactual explanations that involve realistic and actionable changes can be used for the purpose of **Algorithmic Recourse** (AR) to help individuals who face adverse outcomes. An example relevant to the Finance and Economics domain is consumer credit: in this context AR can be used to guide individuals in improving their creditworthiness, should they have previously been denied access to credit based on an automated decision-making system.
@@ -10,4 +12,4 @@ These dynamics may be problematic. As the decision boundary moves in the directi
1012

1113
To the best of my knowledge this is the first work investigating endogenous dynamics in AR. Through future experiments I want to investigate how this phenomenon plays out across different benchmark datasets including German credit, Boston Housing and COMPAS.^[These benchmark datasets have their issues and controversies, which is one of the challenges I would like to discuss at AIES.] Furthermore, I want to assess to what extent the magnitude and direction of domain and model shifts depends on the choice of the counterfactual generator. To this end, I am currently supervising a group of undergraduate students, who are tackling some of these tasks in their final-year research project.
1214

13-
![Dynamics in Algorithmic Recourse: we have a simple Bayesian model trained for binary classification (a); the implementation of AR for a random subset of individuals leads to a domain shift (b); as the classifier is retrained we observe a model shift (c); as this process is repeated, the decision boundary moves away from the target class (d).](poc.png){#fig-dynamics fig.pos="h" width=45%}
15+
![Dynamics in Algorithmic Recourse: we have a simple Bayesian model trained for binary classification (a); the implementation of AR for a random subset of individuals leads to a domain shift (b); as the classifier is retrained we observe a model shift (c); as this process is repeated, the decision boundary moves away from the target class (d).](www/poc.png){#fig-dynamics fig.pos="h" width=45%}

‎paper/tex/before_body.tex‎

100755100644
Lines changed: 36 additions & 17 deletions
Original file line numberDiff line numberDiff line change
@@ -1,17 +1,36 @@
1-
% Include before body: ----
2-
\pagenumbering{gobble}
3-
4-
\newenvironment{acknowledgements}%
5-
{\begin{center}%
6-
\bfseries Acknowledgements
7-
\end{center}}%
8-
{\vfill\null}
9-
\begin{acknowledgements}
10-
...
11-
\end{acknowledgements}
12-
{\vfill\null}
13-
\textbf{DISCLAIMER:}
14-
This is a work-in-progress. Please do not distribute or share.
15-
\pagebreak
16-
17-
\pagenumbering{arabic}
1+
\author{\IEEEauthorblockN{1\textsuperscript{st} Given Name Surname}
2+
\IEEEauthorblockA{\textit{dept. name of organization (of Aff.)} \\
3+
\textit{name of organization (of Aff.)}\\
4+
City, Country \\
5+
email address or ORCID}
6+
\and
7+
\IEEEauthorblockN{2\textsuperscript{nd} Given Name Surname}
8+
\IEEEauthorblockA{\textit{dept. name of organization (of Aff.)} \\
9+
\textit{name of organization (of Aff.)}\\
10+
City, Country \\
11+
email address or ORCID}
12+
\and
13+
\IEEEauthorblockN{3\textsuperscript{rd} Given Name Surname}
14+
\IEEEauthorblockA{\textit{dept. name of organization (of Aff.)} \\
15+
\textit{name of organization (of Aff.)}\\
16+
City, Country \\
17+
email address or ORCID}
18+
\and
19+
\IEEEauthorblockN{4\textsuperscript{th} Given Name Surname}
20+
\IEEEauthorblockA{\textit{dept. name of organization (of Aff.)} \\
21+
\textit{name of organization (of Aff.)}\\
22+
City, Country \\
23+
email address or ORCID}
24+
\and
25+
\IEEEauthorblockN{5\textsuperscript{th} Given Name Surname}
26+
\IEEEauthorblockA{\textit{dept. name of organization (of Aff.)} \\
27+
\textit{name of organization (of Aff.)}\\
28+
City, Country \\
29+
email address or ORCID}
30+
\and
31+
\IEEEauthorblockN{6\textsuperscript{th} Given Name Surname}
32+
\IEEEauthorblockA{\textit{dept. name of organization (of Aff.)} \\
33+
\textit{name of organization (of Aff.)}\\
34+
City, Country \\
35+
email address or ORCID}
36+
}

‎paper/tex/preamble.tex‎

Lines changed: 8 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,8 @@
1-
% Packages:
2-
\usepackage{placeins}
3-
\usepackage{hyperref}
1+
\usepackage{cite}
2+
\usepackage{amsmath,amssymb,amsfonts}
3+
\usepackage{algorithmic}
4+
\usepackage{graphicx}
5+
\usepackage{textcomp}
6+
\usepackage{xcolor}
7+
\def\BibTeX{{\rm B\kern-.05em{\sc i\kern-.025em b}\kern-.08em
8+
T\kern-.1667em\lower.7ex\hbox{E}\kern-.125emX}}

‎paper/www/cat.png‎

16.6 KB
Loading

0 commit comments

Comments
 (0)