Название: New Kind of Machine Learning–Cellular Automata Model
Автор: Parimal Pal Chaudhuri, Adip Dutta, Somshubhro Pal Choudhury, Dipanwita Roy Chowdhury, Raju Hazari
Издательство: Springer
Год: 2025
Страниц: 407
Язык: английский
Формат: pdf (true), epub
Размер: 38.0 MB

This book introduces the CAML model, a novel integration of Cellular Automata (CA) and Machine Learning (ML), designed to deliver efficient computation with minimal training data and low computing resources. CAML operates through two key perspectives: one where CA is enhanced by ML to handle complex non-linear evolution, and another where CA strengthens ML by leveraging linear CA evolution to process linear functions effectively.

The book focuses on real-world applications of CA, such as in Computational Biology, where CAML models protein chains to predict mutations linked to human diseases, using carefully designed CA rule sequences for each amino acid. Another significant application is in multi-language Sentiment Analysis, where the model analyzes text in five languages (Hindi, Arabic, English, Greek, and Georgian), without relying on pre-trained language models.

The input data, used in the experiment, is generated using Python library Scikit-learn and are defined as n-bit binary strings. Certain linear transformations are performed and with different noise in the data generated to get the desired binary dataset required for the experimentation in the system. To use dataset, available in public domain, it is required to create binary data to be used for the experimental setup in the current CA model.

CAML uses CA rules for Unicode character modeling, offering a transparent, interpretable prediction algorithm.

Overall, CAML aims to drive industrial and societal applications of CA, with an emphasis on transparent results and efficient hardware design through CA’s regular, modular, and scalable structure.