Skip to content

Commit ecee71b

Browse files
committed
wiml
1 parent b4eba17 commit ecee71b

File tree

12 files changed

+149
-71
lines changed

12 files changed

+149
-71
lines changed

_bibliography/references.bib

Lines changed: 12 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,17 @@
11
---
22
---
3+
@inproceedings{
4+
rubio-madrigal2025fixed,
5+
title={Fixed Aggregation Features Can Rival {GNN}s},
6+
author={Celia Rubio-Madrigal and Rebekka Burkholz},
7+
booktitle={Women in Machine Learning Workshop @ NeurIPS 2025},
8+
year={2025},
9+
url={https://openreview.net/forum?id=OHgWEMce80},
10+
pdf={https://openreview.net/pdf?id=OHgWEMce80},
11+
img={fixed-aggregation-features.png},
12+
abstract={Graph neural networks (GNNs) are widely believed to excel at node representation learning through trainable neighborhood aggregations. We challenge this view by introducing Fixed Aggregation Features (FAFs), a training-free approach that transforms graph learning tasks into tabular problems. This simple shift enables the use of well-established tabular methods, offering strong interpretability and the flexibility to deploy diverse classifiers. Across 14 benchmarks, well-tuned multilayer perceptrons trained on FAFs rival or outperform state-of-the-art GNNs and graph transformers on 12 tasks -- often using only mean aggregation. The only exceptions are the Roman Empire and Minesweeper datasets, which typically require unusually deep GNNs. To explain the theoretical possibility of non-trainable aggregations, we connect our findings to Kolmogorov–Arnold representations and discuss when mean aggregation can be sufficient. In conclusion, our results call for (i) richer benchmarks benefiting from learning diverse neighborhood aggregations, (ii) strong tabular baselines as standard, and (iii) employing and advancing tabular models for graph data to gain new insights into related tasks.},
13+
}
14+
315
@inproceedings{ zhou2025payattentionsmallweights,
416
title={Pay Attention to Small Weights},
517
author={Chao Zhou and Tom Jacobs and Advait Gadhikar and Rebekka Burkholz},

_data/alumni_members.yml

Lines changed: 17 additions & 15 deletions
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,17 @@
11

2+
- role: Research scientists
3+
full_width: true
4+
members:
5+
- name: Nikita Vedeneev
6+
last_name: Vedeneev
7+
photo: c01mive.jpg
8+
start_date: Dec 24
9+
end_date: May 25
10+
email: mikita.vedzeneyeu@cispa.de
11+
url: https://github.com/nikitaved
12+
description: "I am interesting in making modern AI models efficient. In particular, I work on discovering and exploiting structure in Neural Networks (sparsity, low-dimensional representations and similar) for efficient training, fine-tuning and inference. I am a former full-time core developer for [PyTorch](https://github.com/pytorch/pytorch) and [Lightning Thunder](https://github.com/Lightning-AI/lightning-thunder). Check my [GitHub](https://github.com/nikitaved) to see what I work on now."
13+
next: Senior Engineer at NVIDIA
14+
215
- role: Research assistants
316
members:
417
- name: Ben Horvath
@@ -14,33 +27,21 @@
1427
end_date: Oct 24
1528
email: adarsh.jamadandi@cispa.de
1629
url: https://adarshmj.github.io
17-
next: PhD candidate at IRISA Rennes
30+
next: PhD at IRISA Rennes
1831

1932
- name: Harsha Nelaturu
2033
last_name: Nelaturu
2134
photo: c02hane.jpg
2235
start_date: Aug 23
2336
end_date: Jul 24
2437
url: https://nelaturuharsha.github.io/
25-
next: Applied Scientist Intern at Amazon
38+
next: PhD at Zuse Institute Berlin
2639

2740
- name: Advait Athreya
2841
last_name: Athreya
2942
start_date: Dec 21
3043
end_date: Oct 22
3144

32-
- role: Research engineers
33-
members:
34-
- name: Nikita Vedeneev
35-
last_name: Vedeneev
36-
photo: c01mive.jpg
37-
start_date: Dec 24
38-
end_date: May 25
39-
email: mikita.vedzeneyeu@cispa.de
40-
url: https://github.com/nikitaved
41-
description: "I am interesting in making modern AI models efficient. In particular, I work on discovering and exploiting structure in Neural Networks (sparsity, low-dimensional representations and similar) for efficient training, fine-tuning and inference. I am a former full-time core developer for [PyTorch](https://github.com/pytorch/pytorch) and [Lightning Thunder](https://github.com/Lightning-AI/lightning-thunder). Check my [GitHub](https://github.com/nikitaved) to see what I work on now."
42-
next: Senior Engineer at NVIDIA
43-
4445
- role: Visiting students
4546
members:
4647
- name: Mohammed Adnan
@@ -53,7 +54,7 @@
5354
last_name: Chaves
5455
from: University of Campinas
5556
start_date: Sep 25
56-
end_date: ongoing
57+
end_date: Nov 25
5758
url: https://scholar.google.com/citations?user=dDfbSWUAAAAJ
5859
- name: Otto Piramuthu
5960
last_name: Piramuthu
@@ -62,6 +63,7 @@
6263
end_date: Aug 24
6364

6465
- role: Past Master's theses
66+
full_width: true
6567
members:
6668
- name: Sree Harsha Nelaturu
6769
thesis: "Enhancing the Performance of Random Sparse Masks for Image Classification via Knowledge Distillation"

_data/outreach.yml

Lines changed: 23 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,26 @@
11
videos:
2+
# - title: "Mask in the Mirror: Implicit Sparsification"
3+
# date: 2025-11-17
4+
# speaker: "Tom Jacobs"
5+
# venue: "ELLIS [DL Math+Efficiency] reading group"
6+
# video: https://www.youtube.com/embed/udKX0Y69O08?si=U2dSnD4vkY5s7ERM
7+
# papers:
8+
# - title: "Mask in the Mirror: Implicit Sparsification"
9+
# authors: "Tom Jacobs, and Rebekka Burkholz"
10+
# conference: ICLR 2025
11+
# link: https://openreview.net/forum?id=U47ymTS3ut
12+
13+
# - title: "GNNs Getting ComFy: Community and Feature Similarity Guided Rewiring"
14+
# date: 2025-10-29
15+
# speaker: "Celia Rubio-Madrigal"
16+
# venue: "GLOW reading group"
17+
# video: https://www.youtube.com/embed/L-EH9iys8gc?si=5m7o6BlwaPOktgCG
18+
# papers:
19+
# - title: "GNNs Getting ComFy: Community and Feature Similarity Guided Rewiring"
20+
# authors: "Celia Rubio-Madrigal*, Adarsh Jamadandi*, and Rebekka Burkholz"
21+
# conference: ICLR 2025
22+
# link: https://openreview.net/forum?id=g6v09VxgFw
23+
224
- title: "Weight Decay Controls Implicit Regularization: Insights on Generalization and Sparsity"
325
date: 2025-08-14
426
speaker: "Tom Jacobs"
@@ -13,6 +35,7 @@ videos:
1335
authors: "Tom Jacobs, Chao Zhou, and Rebekka Burkholz"
1436
conference: ICML 2025
1537
link: https://openreview.net/forum?id=MLiR9LS5PW
38+
1639
- title: "Rewiring Graph Neural Networks: When Less is More and Structure Matters"
1740
date: 2025-02-13
1841
speaker: "Celia Rubio-Madrigal"

_pages/home.md

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -8,8 +8,8 @@ permalink: /
88

99
# Relational Machine Learning Lab
1010

11-
> **<span style="color:#d9534f">New positions available!</span>**<br>
12-
> We are currently hiring for PhD and Postdoc positions. Check the [openings](/openings) page and apply now!
11+
> **Open positions — extended deadline**
12+
> We are recruiting PhD candidates and postdoctoral researchers. See the <a href="/openings"><strong style="color:#d9534f">Openings</strong></a> page and apply now!
1313
1414
Welcome! We are the Relational ML research group.
1515
We are part of the [CISPA Helmholtz Center for Information Security](https://cispa.de) in Saarbrücken and St. Ingbert, Germany, and are grateful to [Saarland University (UdS)](https://www.uni-saarland.de) for granting us supervision rights.

_pages/openings.md

Lines changed: 8 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -8,8 +8,8 @@ permalink: /openings
88

99
# Open positions
1010

11-
> **<span style="color:#d9534f">New positions available!</span>**<br>
12-
> We are currently hiring for PhD and Postdoc positions. Check the details below and apply now!
11+
> **Open positions — extended deadline**
12+
> We are recruiting PhD candidates and postdoctoral researchers. We will consider candidates until positions are filled — please apply via the [CISPA Career portal](https://career.cispa.de/jobs/group-relationalml-53).
1313
1414
Are you curious about some of our research and have further questions? Feel free to drop us an e-mail ([burkholz@cispa.de](mailto:burkholz@cispa.de)) to get in touch.
1515

@@ -49,11 +49,14 @@ We are a small team with a flat management structure and a collaborative work cu
4949

5050
The starting dates of the positions are flexible. We are committed to providing a healthy work environment and fostering diversity and respectful interaction. We welcome applications by candidates from all backgrounds and also support non-standard careers.
5151

52+
### Current open positions
53+
54+
* We have PhD and postdoc positions available for 2026.
55+
* [PhD and Postdocs in Efficient Deep Learning](https://career.cispa.de/jobs/group-relationalml-53) at CISPA Helmholtz Center for Information Security.
56+
5257
### Past open positions
58+
This is a non-exhaustive list of past open positions in our group.
5359

54-
* We have some PhD and postdoc positions available for 2026.
55-
* [PhD and Postdocs in Efficient Deep Learning](https://career.cispa.de/jobs/group-relationalml-53)
56-
* [Postdoctoral Researcher in Efficient Deep Learning](https://de.linkedin.com/jobs/view/postdoctoral-researcher-in-efficient-deep-learning-at-relational-machine-learning-lab-4300176253).
5760
* We received an ERC Starting Grant in 2023 ([SPARSE-ML](https://cispa.de/en/research/grants/sparse-ml)) and had several open positions for PhD students and Postdocs:
5861
* [PhD position in sparse machine learning](https://euraxess.ec.europa.eu/jobs/144401).
5962
* [Postdoctoral position in sparse machine learning](https://euraxess.ec.europa.eu/jobs/144392).

_pages/team.md

Lines changed: 5 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -62,10 +62,7 @@ permalink: /team/
6262
{% assign number_printed = 0 %}
6363
{% for group in site.data.alumni_members %}
6464

65-
{% assign full_width = true %}
66-
<!-- {% if group.role contains "Research assistants" or group.role contains "Master's thesis" %}
67-
{% assign full_width = true %}
68-
{% endif %} -->
65+
{% assign full_width = group.full_width %}
6966

7067
{% assign mod_two = number_printed | modulo: 2 %}
7168
{% if full_width and mod_two != 0 %}
@@ -87,11 +84,11 @@ permalink: /team/
8784

8885
<h3>{{group.role}}</h3>
8986
{% for member in group.members %}
90-
{% if member.url %}<a href="{{ member.url }}">{{ member.name }}</a>{% else %}{{ member.name }}{% endif -%}
87+
{% if member.start_date %}{{ member.start_date }}-{{ member.end_date }}: {% endif %}{% if member.url %}<a href="{{ member.url }}">{{ member.name }}</a>{% else %}{{ member.name }}{% endif -%}
9188
{%- if member.from %} ({{ member.from }}){%- endif -%}
92-
{%- if member.thesis -%}: <i>{% if member.url_thesis -%}<a href="{{ member.url_thesis }}">{{ member.thesis }}</a>{% else %}{{ member.thesis }}{% endif %}</i>{%- if member.date_thesis -%}, {{ member.date_thesis }}{%- endif -%}.{%- else -%}:
93-
{{ member.start_date }}-{{ member.end_date }}{%- if member.next -%}. Next {%- endif -%}
94-
{% for next in member.next %} ⇢ {{ next }}{% endfor %}.{%- endif %}
89+
{%- if member.thesis -%}: <i>{% if member.url_thesis -%}<a href="{{ member.url_thesis }}">{{ member.thesis }}</a>{% else %}{{ member.thesis }}{% endif %}</i>{%- if member.date_thesis -%}, {{ member.date_thesis }}{%- endif -%}.{%- else -%}
90+
{%- if member.next -%} {%- endif -%}
91+
{% for next in member.next %} ⇢ {{ next }}{% endfor %}{%- endif %}
9592
{% endfor %}
9693
</div>
9794

182 KB
Loading

_site/index.html

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -111,8 +111,8 @@
111111
<h1 id="relational-machine-learning-lab">Relational Machine Learning Lab</h1>
112112

113113
<blockquote>
114-
<p><strong><span style="color:#d9534f">New positions available!</span></strong><br />
115-
We are currently hiring for PhD and Postdoc positions. Check the <a href="/openings">openings</a> page and apply now!</p>
114+
<p><strong>Open positions — extended deadline</strong><br />
115+
We are recruiting PhD candidates and postdoctoral researchers. See the <a href="/openings"><strong style="color:#d9534f">Openings</strong></a> page and apply now!</p>
116116
</blockquote>
117117

118118
<p>Welcome! We are the Relational ML research group.

_site/openings.html

Lines changed: 11 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -76,8 +76,8 @@
7676
<h1 id="open-positions">Open positions</h1>
7777

7878
<blockquote>
79-
<p><strong><span style="color:#d9534f">New positions available!</span></strong><br />
80-
We are currently hiring for PhD and Postdoc positions. Check the details below and apply now!</p>
79+
<p><strong>Open positions — extended deadline</strong><br />
80+
We are recruiting PhD candidates and postdoctoral researchers. We will consider candidates until positions are filled — please apply via the <a href="https://career.cispa.de/jobs/group-relationalml-53">CISPA Career portal</a>.</p>
8181
</blockquote>
8282

8383
<p>Are you curious about some of our research and have further questions? Feel free to drop us an e-mail (<a href="mailto:burkholz@cispa.de">burkholz@cispa.de</a>) to get in touch.</p>
@@ -125,15 +125,20 @@ <h3 id="what-we-offer">What we offer</h3>
125125

126126
<p>The starting dates of the positions are flexible. We are committed to providing a healthy work environment and fostering diversity and respectful interaction. We welcome applications by candidates from all backgrounds and also support non-standard careers.</p>
127127

128-
<h3 id="past-open-positions">Past open positions</h3>
128+
<h3 id="current-open-positions">Current open positions</h3>
129129

130130
<ul>
131-
<li>We have some PhD and postdoc positions available for 2026.
131+
<li>We have PhD and postdoc positions available for 2026.
132132
<ul>
133-
<li><a href="https://career.cispa.de/jobs/group-relationalml-53">PhD and Postdocs in Efficient Deep Learning</a></li>
134-
<li><a href="https://de.linkedin.com/jobs/view/postdoctoral-researcher-in-efficient-deep-learning-at-relational-machine-learning-lab-4300176253">Postdoctoral Researcher in Efficient Deep Learning</a>.</li>
133+
<li><a href="https://career.cispa.de/jobs/group-relationalml-53">PhD and Postdocs in Efficient Deep Learning</a> at CISPA Helmholtz Center for Information Security.</li>
135134
</ul>
136135
</li>
136+
</ul>
137+
138+
<h3 id="past-open-positions">Past open positions</h3>
139+
<p>This is a non-exhaustive list of past open positions in our group.</p>
140+
141+
<ul>
137142
<li>We received an ERC Starting Grant in 2023 (<a href="https://cispa.de/en/research/grants/sparse-ml">SPARSE-ML</a>) and had several open positions for PhD students and Postdocs:
138143
<ul>
139144
<li><a href="https://euraxess.ec.europa.eu/jobs/144401">PhD position in sparse machine learning</a>.</li>

_site/publications/index.html

Lines changed: 54 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -95,6 +95,60 @@ <h2 id="accepted-papers">Accepted papers</h2>
9595
<ol class="bibliography"><li><!-- _layouts/bib.html -->
9696

9797

98+
<div class="row" style="margin-top: 30px;" id="bib-rubio-madrigal2025fixed" authors="Rubio-MadrigalBurkholz">
99+
<div class="preview"><img class="preview z-depth-1 rounded" src="/images/publications/fixed-aggregation-features.png" />
100+
<!-- include figure.html
101+
path=entry_path
102+
class="preview z-depth-1 rounded"
103+
zoomable=false
104+
alt=entry.img --></div>
105+
<!-- Entry bib key -->
106+
<div id="rubio-madrigal2025fixed" class="col-sm-10">
107+
<!-- Title -->
108+
109+
<div class="title"><a href="https://openreview.net/forum?id=OHgWEMce80"><b>Fixed Aggregation Features Can Rival GNNs</b></a></div>
110+
111+
<!-- Author -->
112+
<div class="author">
113+
114+
115+
Celia Rubio-Madrigal,&nbsp;and&nbsp;Rebekka Burkholz</div>
116+
117+
<!-- Journal/Book title and date -->
118+
119+
120+
<div class="periodical">
121+
<em>In Women in Machine Learning Workshop @ NeurIPS 2025</em>, 2025
122+
</div>
123+
<div class="periodical">
124+
125+
</div>
126+
127+
<!-- Links/Buttons -->
128+
<div class="links"><a class="bibtex btn btn-sm z-depth-0" role="button">Bib</a>
129+
<a class="abstract btn btn-sm z-depth-0" role="button">Abs</a>
130+
<!-- <a href="https://openreview.net/pdf?id=OHgWEMce80" class="btn btn-sm z-depth-0" role="button">PDF</a> -->
131+
</div>
132+
133+
134+
<!-- Hidden abstract block -->
135+
<div class="abstract hidden">
136+
<p>Graph neural networks (GNNs) are widely believed to excel at node representation learning through trainable neighborhood aggregations. We challenge this view by introducing Fixed Aggregation Features (FAFs), a training-free approach that transforms graph learning tasks into tabular problems. This simple shift enables the use of well-established tabular methods, offering strong interpretability and the flexibility to deploy diverse classifiers. Across 14 benchmarks, well-tuned multilayer perceptrons trained on FAFs rival or outperform state-of-the-art GNNs and graph transformers on 12 tasks – often using only mean aggregation. The only exceptions are the Roman Empire and Minesweeper datasets, which typically require unusually deep GNNs. To explain the theoretical possibility of non-trainable aggregations, we connect our findings to Kolmogorov–Arnold representations and discuss when mean aggregation can be sufficient. In conclusion, our results call for (i) richer benchmarks benefiting from learning diverse neighborhood aggregations, (ii) strong tabular baselines as standard, and (iii) employing and advancing tabular models for graph data to gain new insights into related tasks.</p>
137+
</div><!-- Hidden bibtex block -->
138+
<div class="bibtex hidden">
139+
<figure class="highlight"><pre><code class="language-bibtex" data-lang="bibtex"><span class="nc">@inproceedings</span><span class="p">{</span><span class="nl">rubio-madrigal2025fixed</span><span class="p">,</span>
140+
<span class="na">title</span> <span class="p">=</span> <span class="s">{Fixed Aggregation Features Can Rival {GNN}s}</span><span class="p">,</span>
141+
<span class="na">author</span> <span class="p">=</span> <span class="s">{Rubio-Madrigal, Celia and Burkholz, Rebekka}</span><span class="p">,</span>
142+
<span class="na">booktitle</span> <span class="p">=</span> <span class="s">{Women in Machine Learning Workshop @ NeurIPS 2025}</span><span class="p">,</span>
143+
<span class="na">year</span> <span class="p">=</span> <span class="s">{2025}</span><span class="p">,</span>
144+
<span class="na">url</span> <span class="p">=</span> <span class="s">{https://openreview.net/forum?id=OHgWEMce80}</span><span class="p">,</span>
145+
<span class="p">}</span></code></pre></figure>
146+
</div>
147+
</div>
148+
</div></li>
149+
<li><!-- _layouts/bib.html -->
150+
151+
98152
<div class="row" style="margin-top: 30px;" id="bib-zhou2025payattentionsmallweights" authors="ZhouJacobsGadhikarBurkholz">
99153
<div class="preview"><img class="preview z-depth-1 rounded" src="/images/publications/smallweights.png" />
100154
<!-- include figure.html

0 commit comments

Comments
 (0)