Get our free extension to see links to code for papers anywhere online!Free add-on: code for papers everywhere!Free add-on: See code for papers anywhere!

Add to Chrome

Add to Firefox

Add to Edge

Dmitrii Gavrilev

ICML 2023 Topological Deep Learning Challenge : Design and Results

Oct 02, 2023
Mathilde Papillon, Mustafa Hajij, Florian Frantzen, Josef Hoppe, Helen Jenne, Johan Mathe, Audun Myers, Theodore Papamarkou, Michael T. Schaub, Ghada Zamzmi, Tolga Birdal, Tamal Dey, Tim Doster, Tegan Emerson, Gurusankar Gopalakrishnan, Devendra Govil, Vincent Grande, Aldo Guzmán-Sáenz, Henry Kvinge, Neal Livesay, Jan Meisner, Soham Mukherjee, Shreyas N. Samaga, Karthikeyan Natesan Ramamurthy, Maneel Reddy Karri, Paul Rosen, Sophia Sanborn, Michael Scholkemper, Robin Walters, Jens Agerberg, Georg Bökman, Sadrodin Barikbin, Claudio Battiloro, Gleb Bazhenov, Guillermo Bernardez, Aiden Brent, Sergio Escalera, Simone Fiorellino, Dmitrii Gavrilev, Mohammed Hassanin, Paul Häusner, Odin Hoff Gardaa, Abdelwahed Khamis, Manuel Lecha, German Magai, Tatiana Malygina, Pavlo Melnyk, Rubén Ballester, Kalyan Nadimpalli, Alexander Nikitin, Abraham Rabinowitz, Alessandro Salatiello, Simone Scardapane, Luca Scofano, Suraj Singh, Jens Sjölund, Pavel Snopov, Indro Spinelli, Lev Telyatnikov, Lucia Testa, Maosheng Yang, Yixiao Yue, Olga Zaghen, Ali Zia, Nina Miolane

Figure 1 for ICML 2023 Topological Deep Learning Challenge : Design and Results

Figure 2 for ICML 2023 Topological Deep Learning Challenge : Design and Results

This paper presents the computational challenge on topological deep learning that was hosted within the ICML 2023 Workshop on Topology and Geometry in Machine Learning. The competition asked participants to provide open-source implementations of topological neural networks from the literature by contributing to the python packages TopoNetX (data processing) and TopoModelX (deep learning). The challenge attracted twenty-eight qualifying submissions in its two-month duration. This paper describes the design of the challenge and summarizes its main findings.

Via

Access Paper or Ask Questions

Anomaly Detection in Networks via Score-Based Generative Models

Jun 27, 2023
Dmitrii Gavrilev, Evgeny Burnaev

Figure 1 for Anomaly Detection in Networks via Score-Based Generative Models

Figure 2 for Anomaly Detection in Networks via Score-Based Generative Models

Figure 3 for Anomaly Detection in Networks via Score-Based Generative Models

Figure 4 for Anomaly Detection in Networks via Score-Based Generative Models

Node outlier detection in attributed graphs is a challenging problem for which there is no method that would work well across different datasets. Motivated by the state-of-the-art results of score-based models in graph generative modeling, we propose to incorporate them into the aforementioned problem. Our method achieves competitive results on small-scale graphs. We provide an empirical analysis of the Dirichlet energy, and show that generative models might struggle to accurately reconstruct it.

* 16 pages, 8 figures, ICML workshop on Structured Probabilistic Inference & Generative Modeling

Via

Access Paper or Ask Questions

High Performance of Gradient Boosting in Binding Affinity Prediction

May 14, 2022
Dmitrii Gavrilev, Nurlybek Amangeldiuly, Sergei Ivanov, Evgeny Burnaev

Figure 1 for High Performance of Gradient Boosting in Binding Affinity Prediction

Figure 2 for High Performance of Gradient Boosting in Binding Affinity Prediction

Figure 3 for High Performance of Gradient Boosting in Binding Affinity Prediction

Figure 4 for High Performance of Gradient Boosting in Binding Affinity Prediction

Prediction of protein-ligand (PL) binding affinity remains the key to drug discovery. Popular approaches in recent years involve graph neural networks (GNNs), which are used to learn the topology and geometry of PL complexes. However, GNNs are computationally heavy and have poor scalability to graph sizes. On the other hand, traditional machine learning (ML) approaches, such as gradient-boosted decision trees (GBDTs), are lightweight yet extremely efficient for tabular data. We propose to use PL interaction features along with PL graph-level features in GBDT. We show that this combination outperforms the existing solutions.

Via

Access Paper or Ask Questions