Performance analysis for Tsetlin Machine
Organizing and reviewing the analysis of the Tsetlin Machine. The analysis covers the convergence of the Tsetlin machine for the three most fundamental digital operations, "one bit" operation, "AND/OR" operation, and "XOR" operation.
In September Associate Professor Lei Jiao contacted me, asking if I could help with organizing and review the analysis of the Tsetlin machine. The original analysis was handwritten, to make it more applicable, the analysis was revised in a digital program.
The analysis covers the convergence of the Tsetlin machine for the three most fundamental digital operations, "one bit" operation, "AND/OR" operation, and "XOR" operation. A Tsetlin Machine solves complex pattern recognition problems with easy-to-interpret propositional formulas composed by many Tsetlin Automata. The analysis provides the reader with Markov Chain illustrations which makes it easier to understand the actual analysis, the Markov Chain is a mathematical system where it hops from one state to another. In this analysis, the states are, include and exclude, where the Tsetlin machine can receive a penalty or reward based on its actions.
During the organization of this analysis, I've achieved a better understanding of Markov Chain and the Tsetlin Machine, from a mathematical point of view. The Tsetlin Machine is a powerful tool which I can make use of at a later time as an alternative to neural networks since in some experiments the Tsetlin Machine has performed better than other alternatives.