This week, I worked on improving the base models for making simulated data. I’ve found that the hyperbolic secant function is a great representation of the peaks in muscle movements, so I’ve been using linear combinations of that to make a good prior distribution function. I have also been familiarizing myself with scikit, a python package that has SVM support, and could help us with our classifier. 

Part of our project involves classification of signals, and to better aid that classification we have been considering making a compound classifier. The original results of our classifier were not close to the accuracy we were hoping for, so to add on to this I have also been researching into Neural Networks and Decision Trees. Along with Support Vector Machines and Maximum a Posteriori, incorporating another classifier into our project could give us a higher classification accuracy. We could take the confidence of the outputs produced by each classifier and weigh them to decide which output to proceed with. We could also use these confidences as a recurrent feature to update and train our classifier system with. 

We are currently working towards our third integration test, and hope to have everything together by Thanksgiving. I have also been in the process of packing, as I will be moving during the break, and have planned out and prepared the parts of my project that I need to take home with me. Once we have a fully integrated product in the coming days, we can work together to start refining our project and making it run smoothly and efficiently.