AKII Finalist in Merck “Future of AI” Challenge
Dr. Axenie together with two students of THI (team NeuroTHIx: Du Xiaorui, Yavuzhan Erdem, Cristian Axenie) qualified for the final of Merck “Future of AI” Challenge.
IRENA (Invariant Representations Extraction in Neural Architectures)
In this project we aim at building an unsupervised learning system that is based on and
inspired by our biological intelligence for the problem of learning invariant representations.
Mammalian visual systems are characterized by their ability to recognize stimuli invariant to
various transformations. With our proposed model, we investigate the hypothesis that this
ability is achieved by the temporal encoding of visual stimuli, and why not, other sensory
stimuli. By using a model of a multisensory cortically inspired network, we show that this
encoding is invariant to several transformations and robust with respect to stimulus
variability. Furthermore, we show that the proposed model provides a rapid encoding and
computation, in accordance with recent physiological results.
The team is now invited in Darmstadt at Merck’s Research Center for a 2 day boot camp to refine the idea and then work on it until end August. To conclude the Challenge, we will present our final solution in Darmstadt on 30th of August.
Well done and good luck!