This article will entail a framework for creating and discovering new bio-inspired algorithms both for machine learning and engineering purposes as well as looking at other potential fields that might offer solutions within a computational model. It is in a way a brainstorm for a possibility.
As you might be familiar, Brain-inspired Computing applies to fields from Brain-inspired computer chips produced by companies like IBM(examples: EU-backed SpiNNaker and BrainScaleS, Stanford’s Neurogrid, IBM’s TrueNorth, and Qualcomm’s Zeroth) to Artificial Intelligence solutions. I want to briefly go over some types of Artificial Intelligence uses of bio-inspired computing. The range is vast. To reiterate a few I can list Artificial Neural Networks(ANNs), Genetic Algorithms, Neuroevolution. What all these concepts are mathematical representations of biological structures or processes that are beneficial for change and/or growth.
Trending AI Articles:
1. Cheat Sheets for AI, Neural Networks, Machine Learning, Deep Learning & Big Data
2. Data Science Simplified Part 1: Principles and Process
3. Getting Started with Building Realtime API Infrastructure
4. Bursting the Jargon bubbles — Deep Learning
If biology offers so many wonderful solutions, what’s next? Are we going to stick with what is given to us or could we make adjustments? Evolution is a trial and error method in a sense. There are evolutionary pressures, there are adaptations, mutations and of course a ton of random noise in addition to mismatches. Mismatches are everywhere within nature and even the human body. Things that we don’t need anymore but are still a part of our bodies are examples of these. We do not need body hair in the modern world but we used to need it well before the industrial revolution and hunter-gatherer ages where humans needed a way to protect themselves from cold. From our pinky toes to our overindulgence in excessive sugar and salt, it’s everywhere.
Human’s are intelligent enough to understand these factors, so could we potentially improve on the biological structures and processes we mathematically use for computational models? The answer is yet to be fully answered. If we can genetically improve humans from their DNA, we should also be able to tweak an algorithm to make it better. While thinking about such a diverse subject, it is essential to go back and forth between biology and computer science. Biology is the base of the inspiration therefor we can examine the problems as well as the gaps that leave room for advancement. Going off of the article that I mentioned in the begging, is sleep actually the most efficient way to improve memory. Structurally humans are bound to sleep. It is one of the most powerful instinct but there could be a better way if we were to take all the knowledge we have and create a structure from scratch. Evolution on the other hand is blind to human values, morality and information. So for evolution it doesn’t matter if a human feels emotional pain or not or feels horrible when sleep deprived. Whatever makes our genes survive and reproduce, that’s what evolution keeps. We could integrate our holistic knowledge and values into our improvements. We could make algorithms that are more time and power efficient going off of the biological model.
We do not only have to look at biology to discover more. There’s physics, chemistry and many more fields that can be a building block for something useful. Hopefully, this theoretical exploration sparks a scientific inspiration.