The following are some of my published works and articles.
You can also find my profile on Google Scholar and Semantic Scholar.
Publications
-
Retrospective for the Dynamic Sensorium Competition for predicting large-scale mouse primary visual cortex activity from videos
Polina Turishcheva, Paul G. Fahey, Michaela Vystrčilová, and 22 more authors
2024
Understanding how biological visual systems process information is challenging because of the nonlinear relationship between visual input and neuronal responses. Artificial neural networks allow computational neuroscientists to create predictive models that connect biological and machine vision. Machine learning has benefited tremendously from benchmarks that compare different model on the same task under standardized conditions. However, there was no standardized benchmark to identify state-of-the-art dynamic models of the mouse visual system. To address this gap, we established the Sensorium 2023 Benchmark Competition with dynamic input, featuring a new large-scale dataset from the primary visual cortex of ten mice. This dataset includes responses from 78,853 neurons to 2 hours of dynamic stimuli per neuron, together with the behavioral measurements such as running speed, pupil dilation, and eye movements. The competition ranked models in two tracks based on predictive performance for neuronal responses on a held-out test set: one focusing on predicting in-domain natural stimuli and another on out-of-distribution (OOD) stimuli to assess model generalization. As part of the NeurIPS 2023 competition track, we received more than 160 model submissions from 22 teams. Several new architectures for predictive models were proposed, and the winning teams improved the previous state-of-the-art model by 50%. Access to the dataset as well as the benchmarking infrastructure will remain online at www.sensorium-competition.net.
-
Knowledge Graph Embeddings in the Biomedical Domain: Are They Useful? A Look at Link Prediction, Rule Learning, and Downstream Polypharmacy Tasks
2023
Knowledge graphs are powerful tools for representing and organising complex biomedical data. Several knowledge graph embedding algorithms have been proposed to learn from and complete knowledge graphs. However, a recent study demonstrates the limited efficacy of these embedding algorithms when applied to biomedical knowledge graphs, raising the question of whether knowledge graph embeddings have limitations in biomedical settings. This study aims to apply state-of-the-art knowledge graph embedding models in the context of a recent biomedical knowledge graph, BioKG, and evaluate their performance and potential downstream uses. We achieve a three-fold improvement in terms of performance based on the HITS@10 score over previous work on the same biomedical knowledge graph. Additionally, we provide interpretable predictions through a rule-based method. We demonstrate that knowledge graph embedding models are applicable in practice by evaluating the best-performing model on four tasks that represent real-life polypharmacy situations. Results suggest that knowledge learnt from large biomedical knowledge graphs can be transferred to such downstream use cases.
-
QMaxSATpb: A Certified MaxSAT Solver
In Logic Programming and Nonmonotonic Reasoning, 2022
While certification has been successful in the context of satisfiablity solving, with most state-of-the-art solvers now able to provide proofs of unsatisfiability, in maximum satisfiability, such techniques are not yet widespread. In this paper, we present QMaxSATpb, an extension of QMaxSAT that can produce proofs of optimality in the VeriPB proof format, which itself builds on the well-known cutting planes proof system. Our experiments demonstrate that proof logging is possible without much overhead.
-
LP2PB: Translating Answer Set Programs into Pseudo-Boolean Theories
In Proceedings of the 36th International Conference on Logic Programming: Technical Communications, 2020
Answer set programming (ASP) is a well-established knowledge representation formalism. Most ASP solvers are based on (extensions of) technology from Boolean satisfiability solving. While these solvers have shown to be very successful in many practical applications, their strength is limited by their underlying proof system, resolution. In this paper, we present a new tool LP2PB that translates ASP programs into pseudo-Boolean theories, for which solvers based on the (stronger) cutting plane proof system exist. We evaluate our tool, and the potential of cutting-plane–based solving for ASP on traditional ASP benchmarks as well as benchmarks from pseudo-Boolean solving. Our results are mixed: overall, traditional ASP solvers still outperform our translational approach, but several benchmark families are identified where the balance shifts the other way, thereby suggesting that further investigation into a stronger proof system for ASP is valuable.
Theses
-
-
Transfer learning in Brain-Computer Interfaces: Language-Pretrained Transformers for Classifying Electroencephalography
2022
(MSc Computer Science, Vrije Universiteit Brussel)