University of Agder Logo

Milestones and discoveries

TMComposites

TMComposites: Plug-and-Play Collaboration Between Specialized Tsetlin Machines

A new paper introduces the concept of Tsetlin Machine Composites. A Tsetlin Machine Composite merges multiple Tsetlin Machines with different specialities, significantly boosting team performance. The architecture is plug-and-play in that the individual Tsetlin Machines can train independently and then combine in any way, at any time, without fine-tuning. The architecture yields new state-of-the-art results for Tsetlin Machines on Fashion-MNIST, CIFAR-10, and CIFAR-100, using three image processing methods for demonstration.

Logical Word Embedding with Tsetlin Machine Autoencoder

Do we really need 300 floats to represent the meaning of a word? Representing words with words - a logical approach to word embedding using a self-supervised Tsetlin Machine Autoencoder.

A new self-supervised machine learning approach captures word meaning with concise logical expressions. The logical expressions consist of contextual words like "black," "cup," and "hot" to define other words like "coffee," thus being human-understandable. This logical embedding performs competitively on several intrinsic and extrinsic benchmarks, matching pre-trained GLoVe embeddings on six downstream classification tasks.

Show more