@emsimons
Visiting Fulbright Student @ the AI for Data-Oriented Science Lab. Lifelong learner. Chronically caffeinated champion of crosswords, comedy, and the Oxford comma. Human with humility. And avid alliteration adopter.
An overview of our edge pooling method.
Big graph, little compute?
In our #NeurIPS2025 paper we propose geometry-aware edge-contraction-based pooling methods for GNNs.
Our methods preserve graph structure, make interpretable pooling choices, and ensure robust performance at downstream tasks.
π§΅1/n
An overview of the Euler Characteristic Transform, illustrated for a single Euler Characteristic Curve, i.e., the calculation of Euler characteristics along a single direction.
#MachineLearning loves data. #Topology loves shape.
Let's bring them together!
πThe Euler Characteristic Transform (ECT) bridges both worlds, as I outline in a recent article in the Notices of the
@amermathsoc.bsky.social
Engage!
π§΅1/n
Join us on a (mathematical) journey to a shire - oops, HIGHER - standard and principled evaluation schema for our benchmark datasets. This is the reward of the RINGS framework.
π Blog: aidos.group/blog/rings/
π Paper: doi.org/10.48550/arX...
π©βπ» Code: github.com/aidos-lab/ri...
β¨Excited to be at ICML 2025 to share our new evaluation framework for graph-learning datasets, work with @dataspider.bsky.social, @jeremy-wayland.bsky.social, @pseudomanifold.topology.rocks
Curious about who shall pass, and who shall not?π€
Enjoy nods to LoTR?π§ββοΈ
Check out our blog and paper below!
New blog post on overcoming academic impostor syndrome. This arose from great discussions during our group retreat, and I decided to make it publicly available in the hope of providing some utility to #academia.
#AcademicSky #AcademicChatter
bastian.rieck.me/blog/2025/im...
A subset of the AIDOS Lab standing in front of a panoramic view of Bern
Our AIDOS Lab website has been fully revamped, thanks to a massive team effort as part of a very fun retreat in #Bern:
aidos.group
Check out our research directions and some of the things we have been up to these days!
#Academia #MachineLearning
Huge congratulations, Bastian! πIncredibly well-deserved. βοΈ
Genuine! π
We hope SCOTT can prove helpful in shepherding your flock of graphs through curvature filtrations.π
Big thanks to @jeremy-wayland.bsky.social for his incredible mentorship throughout. Special thanks also to our Scotland consultants: Mairi McClean and @limbeckkat.bsky.social.π΄σ §σ ’σ ³σ £σ ΄σ Ώ
βTil next time!
π§΅7/n
It is also easy to choose your own SCOTT-ish adventure.π
Tailor your KILTπͺ‘ with customizable curvature measures:
πForman & Balanced Forman
πOllivier-Ricci
πResistance
And vectorize by either:
ποΈ outlining a persistence landscape, or
πΈ snapping a persistence image
π§΅6/n
The meat of it:
π©βπ³ curvature computations with a KILT (Krvature-Informed Links and Topology) object
π©βπ³ persistent homology via edge filtrations
π©βπ³ vectorization of persistence diagrams
π©βπ³ distances between vectorizations
β¦all baked into a digestible SCOTT-ish haggis.π½οΈ
π§΅5/n
By combining discrete curvature on graphs with persistent homology, SCOTT provides expressive descriptors of graph sets that are:
β¨Robust
β¨Stable
β¨Expressive
β¨Compatible with Statistical Testing
β¦all made easier - and breezier than a day in the Scottish Highlands.π¨
π§΅4/n
Via SCOTT, these methods are now pip installable on a venv near you! π
π»: github.com/aidos-lab/cu...
π: aidos.group/curvature-fi...
π§΅3/n
π Developed with @jeremy-wayland.bsky.social, @pseudomanifold.topology.rocks, SCOTT is the new code base supporting their paper with @joshsouthern.bsky.social @mmbronstein.bsky.social.
Check out the paper here:
π: arxiv.org/abs/2301.12906
π§΅2/n
scott-logo
Hi! π Emily Simons here, Fulbright Student Researcher with the AIDOS Lab at #HelmholtzMunich.
Happy to be sharing my first contribution to AIDOS in this π§΅.
Say hiya to SCOTT, the perfect holiday (software) packageπfor the #curvature and #graph enthusiast in your life.
π§΅1/n