Amazon Web Services partner and AI specialist Max Kelsen has been featured in arguably the premier scientific academic publication for one its machine learning research projects.
The study, titled “Verifying explainability of a deep learning tissue classifier trained on RNA-seq data”, was published in Nature Scientific Reports last week.
In essence, the research aims to show that deep learning is applicable to a viable use case in healthcare and is also more robust and more useful than traditional statistical techniques.
It also aims to show that deep learning models are fundamentally explainable, and those explanations are reliable, robust, and what would be expected, and eventually make its way into health clinics.
Joining Max Kelsen staff in the research is Brisbane-based QIMR Berghofer Medical Research Institute.
“Explainability is an absolutely critical component of any [machine learning] model built to be used in the healthcare domain. Fantastic achievement to the Max Kelsen Research Lab - Genomics team for delivering great scientific work published in Nature Scientific Reports, in collaboration with our academic friends at QIMR Berghofer Medical Research Institute,” Max Kelsen’s LinkedIn post read.
“Together, we've been able to show the benefits of explainable neural networks over traditional bioinformatic tools when applied to biological multi-class problems.”
Max Kelsen chief executive Nicholas Therkelsen-Terry told CRN it was the company’s first time being published in Nature, which had been a long-term goal for the staff.
“When our research lead Maciej Trzaskowski joined the company he asked us, ‘What’s the goal of this lab and where do you want it to be?’ and we said we would like to be producing research that’s worthy of an actual publication,” Therkelsen-Terry said.
“For us, [being published] is significant, because Nature publishes tonnes of work every day but rarely does that work come from commercial organisations and even rarer, is that coming from a small Australian commercial organisation.”
Cited in the research includes Trzaskowski and other Max Kelsen employees Melvyn Yap, Rebecca L. Johnston, Helena Foley, Samual MacDonald, Khoa Tran and Cameron Bean. They were joined by QIMR Berghofer researchers Olga Kondrashova, Katia Nones, Lambros T. Koufariotis, John V. Pearson and Nicola Waddell.