Kolmogorov-Arnold Networks for Genomic Tasks DOI Creative Commons
Oleksandr Cherednichenko, Maria Poptsova

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 11, 2024

Abstract Kolmogorov-Arnold Networks (KANs) emerged as a promising alternative for multilayer perceptrons in dense fully connected networks. Multiple attempts have been made to integrate KANs into various deep learning architectures the domains of computer vision and natural language processing. Integrating models genomic tasks has not explored. Here, we tested linear (LKANs) convolutional (CKANs) replacement MLP baseline classification generation sequences. We used three benchmark datasets: Genomic Benchmarks, Genome Understanding Evaluation, Flipon Benchmark. demonstrated that LKANs outperformed both CK-ANs on almost all datasets. CKANs can achieve comparable results but struggle with scaling over large number parameters. Ablation analysis KAN layers correlates model performance. Overall, show improving performance relatively small Unleashing potential different SOTA currently genomics requires further research.

Language: Английский

Uncertainty-aware genomic deep learning with knowledge distillation DOI Creative Commons
Jessica Zhou, Kaeli Rizzo, Ziqi Tang

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 15, 2024

Deep neural networks (DNNs) have advanced predictive modeling for regulatory genomics, but challenges remain in ensuring the reliability of their predictions and understanding key factors behind decision making. Here we introduce DEGU (Distilling Ensembles Genomic Uncertainty-aware models), a method that integrates ensemble learning knowledge distillation to improve robustness explainability DNN predictions. distills an DNNs into single model, capturing both average ensemble's variability across them, with latter representing epistemic (or model-based) uncertainty. also includes optional auxiliary task estimate aleatoric, or data-based, uncertainty by experimental replicates. By applying various functional genomic prediction tasks, demonstrate DEGU-trained models inherit performance benefits ensembles improved generalization out-of-distribution sequences more consistent explanations cis-regulatory mechanisms through attribution analysis. Moreover, provide calibrated estimates, conformal offering coverage guarantees under minimal assumptions. Overall, paves way robust trustworthy applications deep genomics research.

Language: Английский

Citations

0
Kolmogorov-Arnold Networks for Genomic Tasks DOI Creative Commons
Oleksandr Cherednichenko, Maria Poptsova

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 11, 2024

Abstract Kolmogorov-Arnold Networks (KANs) emerged as a promising alternative for multilayer perceptrons in dense fully connected networks. Multiple attempts have been made to integrate KANs into various deep learning architectures the domains of computer vision and natural language processing. Integrating models genomic tasks has not explored. Here, we tested linear (LKANs) convolutional (CKANs) replacement MLP baseline classification generation sequences. We used three benchmark datasets: Genomic Benchmarks, Genome Understanding Evaluation, Flipon Benchmark. demonstrated that LKANs outperformed both CK-ANs on almost all datasets. CKANs can achieve comparable results but struggle with scaling over large number parameters. Ablation analysis KAN layers correlates model performance. Overall, show improving performance relatively small Unleashing potential different SOTA currently genomics requires further research.

Language: Английский

Citations

0