Best Tsetlin Machine student paper research at ISTM 2024

Given it sits at the core of our novel approach to edge AI, we were more than pleased for the opportunity to attend the 3rd International Symposium on the Tsetlin Machine in Pittsburgh. But we were absolutely chuffed to sponsor the symposium’s ‘Best Student Paper Award’. Within the world of AI, research typically follows hype, which is to say that the vast majority of students and researchers presently spend their time focussed on generative AI and neural networks. But Tsetlin Machines open up whole new potentials for AI and ML, particularly at the edge.

It was the domain of image processing that took out this year’s student award, with Tsetlin Machines being used with process and benchmark the CIFAR-10 dataset. This has been a challenge tackled in the past. In 2020, the accuracy for Tsetlin Machines processing CIFAR-10 sat at around 61%. That rose to 75.1% in 2023. Neither are ideal. Through the work of the paper’s authors - Ylva Grønningsæter, Halvor S. Smørvik, and Ole-Christoffer Granmo - this year's ISTM saw a new toolbox developed that provides state-of-the-art results on CIFAR-10 for Tsetlin Machines. Accuracy for the dataset now reaches as much as 82.8%.

The code associated with the paper is available from the authors on GitHub while the paper can be read here.

Ylva Grønningsæter presents the winning research paper at ISTM 2024