Patterns Paper Shows Benefits Of Modeling Complicated Small Datasets Using Unconventional, Quantum Computing-Inspired Algorithms

Patterns Paper Shows Benefits of Modeling Complicated Small Datasets Using Unconventional_ Quantum Computing-inspired Algorithms

Genuity Science, a U.S.-headquartered genomics data, analytics and insights organization, today announced that the paper “Quantum processor-inspired machine learning in the biomedical sciences,” was published in the Cell Press journal, Patterns.

The paper discusses the evaluation of special algorithms, known as Ising-type algorithms, which were applied to actual human tumor data from The Cancer Genome Atlas.  These algorithms were inspired by recent developments in physical quantum processors and are relatively unused in the biomedical sciences.

The results of the research show competitive performance of Ising-type algorithms with conventional machine learning algorithms in classifying human cancer types and associated molecular subtypes when training with all available data and superior classification performance over standard machine-learning approaches when used to classify and model small, complex datasets.

This research suggests potential application for rare diseases or other clinical applications where the number of samples may be quite small.

Daniel Lidar, PhD, Viterbi Professor of Engineering at the University of Southern California said “Quantum machine learning is one of the most promising applications of quantum computing. The outside-the-box thinking it involves has yielded many new insights into traditional classical machine learning. Our work demonstrates the power of this dual approach, and it is particularly gratifying to see that we can now tackle datasets of medical relevance using today’s quantum processors.”

Also Read: IoT-Centric Cyber-attacks are Rapidly Increasing

Tom Chittenden, PhD, DPhil, PStat and Chief Technology Officer of Genuity Science said, “These proposals have generated interest in the scientific community and in the general public for their potential to address computationally difficult tasks and to model more complicated data distributions.

This approach for small experimental designs is particularly useful in medicine, where large datasets may be prohibitively expensive to obtain, assessing drug efficacy in clinical trials, or when studying rare diseases.

As technology improves and new algorithms are introduced, we see the potential that unconventional classification algorithms can offer in terms of unique insights and the discovery of novel approaches for solving complex biological challenges.”

For more such updates follow us on Google News ITsecuritywire News.