DeeL@BiCi: Deep Learning: Theory, Algorithms, and Applications
Sprache des Tagungstitel:
Everyday we are exposed to various chemicals via food additives, cleaning, and cosmetic products - and some of them might be toxic.
However testing the toxicity of compounds in such products and drug candidates by biological experiments is neither financially nor logistically feasible.
Therefore the government agencies NIH, EPA and FDA launched the "Tox21 Data Challenge" to assess the performance of computational methods in predicting the toxicity of chemical compounds.
Though deep networks were never applied to tox prediction, they clearly outperformed all other participating methods.
Their strength is that they automatically learn features that correspond to well-established toxicophores but in most cases they construct new ones.
Our deep learning approach won 9 out of 15 challenges including both panel-challenges (nuclear receptors and stress response) as well as the overall Grand Challenge.
Deep learning set a new standard in tox prediction.
Sprache der Kurzfassung:
Hauptvortrag / Eingeladener Vortrag auf einer Tagung