A New Year and BIG dreams and a teeny tiny project I’m doing right now. My present goal is to train my Deep Learning model on a larger number of images believing that the predictions it makes will be more accurate. I’m taking pictures of Pseudopithomyces Chartarum spores as well as other things that aren’t but look similar, with my trinocular microscope.

I’ve created a new Google Colaboratory notebook and I’m in the process of writing the code ready for when I have 5,000 images in my train/validation data set. At this stage it’s a binary classification problem. Is this a Pseudopithomyces Chartarum spore or not? There are some strains apparently that don’t produce spores containing the toxin sporIdesmin. Check out this YouTube video featuring Dr Axel Heiser gives an excellent overview of Facial Eczema research in New Zealand.

Do these non toxic spores look any different? If so then the problem becomes a multi-classification one.

This is my little first step, using A.I. to help automate spore counting in New Zealand. Our country won the 36th America’s Cup in 2021 with the help of a simulator and a deep learning neural network algorithm or ‘bot’ to help run it. It made the testing of boat designs so much faster. A simulator is essentially a game. With an environment, an agent or agents and some kind of reward system.

In the same way a cow or herd of cows could be viewed as an ‘agent’ living within an environment. A multitude of different factors come into play every season as they become pregnant, dry off in the autumn and give birth again in the spring. They impact her production, health and well being. A cow that doesn’t get in calf leaves the herd. It’s a waste especially if she is a young cow. The farmer hasn’t recouped the money invested in her in those first two or three years.

My big dream is to gather a team and funds to create a game based on a simulator of a ‘typical’ dairy farm. Then we’ll craft a deep learning neural network artificial intelligence algorithm like McKearnsey Quantum Black did to run it. The aim will be to win a series of 10 races…one for every season or year. We don’t have to win every race just the series.