Image: (private)
Oh well, “so bad” was not their scientific work, say Johannes Schimunek and Philipp Seidl, both doctoral students at the LIT AI Lab and Institute for Machine Learning at the JKU. And so the two stack enormously deep. Because it was her essay that paved the way for them to go to Africa, to Rwanda, to one of the largest and most important international conferences on the subject of artificial intelligence (AI). “That was a really big thing”sums up Schimunek.
Even before the two were even invited to the fair, they – like hundreds of other scientists – had to write an article about their scientific work. A jury then selected the best submissions and invited their authors. In Rwanda they got the chance to present their work. “But there was a lot of luck involved in the selection”, says Schimunek. And again he piles deep.
The artificial intelligence helps
But what is the work of the two about? The title is “Context-enriched molecule representations improve few-shot drug discovery”. “Now that sounds quite complicated, but it can be explained well”, says Schimunek. Essentially, it is about researching active substances and medicines, about the interaction of molecules and cells. Laboratories are investigating what effects certain molecules could have on cells; the positive effects are of interest here. Since chemists have billions of molecules at their disposal, “but no one has the capacity to try it all”the two programmed an AI.
It is intended to predict which molecules have a positive effect. In order to speed up this process, the molecules are not tested individually, but placed in a context with one another. “This can perhaps be compared to words. These must also always be seen in connection with other words before the content of a sentence can be predicted”the researchers explain. “For example, the word ‘it’ alone is not very meaningful. It must first be searched for a word, i.e. context, to which it refers, e.g. B. ‘the cat'”. Only when such a pairing can be found can one imagine the effect. The effect of the molecules found in this way is then examined in a laboratory.
“Almost like at school”
The two were then allowed to present their results at the multi-day conference – together with other scientists in a large hall. “It was almost like at school, you stand there with your poster and explain to the others what it’s about”, says Schimunek. Seidl and Schimunek had two hours to explain their research.
“I haven’t actually seen that much of Rwanda”, sums up the 31-year-old. He would have spent most of his time at the science fair. The first time he did “sniffed African air” and he was surprised by Rwanda: “It really is a beautiful, green and cool country.” The scientists even went on a small safari.
Source: Nachrichten
