Introuducing The-Ben-3 recognition device.
Using Logistic Regression to train data from an accelerometer in order to recognize and categorize gestures.
Key features: Using simple components like a microcontroller and deploying complex machine learning algorithms to obtain clean results.
All the major components of code are soft coded and hence, you can use them to your advantage, the only "code" you will have to modify, is labeling your data in reshapeDataNormalize.
The objective is using an advanced optimization method in order to minimize the cost function of the logistic Regression algorithm. Then compute Accuracy.
There is an inbuilt function that I have created in order to check for broken data samples, so you dont have to go looking through thousands of lines of code.
testStuff is just a file for you to, well, test any code you are unfamiliar with.
Currently, I am trying to gain decent results by using neural networks, I am having trouble debugging the code, hopefully it will be complete in a few days time.
Then, I will also see the outcome of Support Vector Machines
All changes will be updated.