Machine learning to detect non-human mouse movements on a webpage

Hi all I need to be able to detect non-human mouse movements on a webpage (any webpage). I can provide a large amount of training data (real human mouse movements on a webpage). This would contain the x, y coordinates and timestamp of each mouse movement. For example: x | y | timestamp 1 | 1 | 1538054356 1 | 2 | 1538054359 2 | 3 | 1538054370 2 | 4 | 1538054392 There would be data like this for thousands of humans. I need the following from you: 1. You figure out what is the best way to train a model for this. You need to document this: both your logic and how to create the model. 2. The model should be able to accept the same input as the training data above (x, y, timestamp). For example, an array of dictionaries of data: [{x: 1, y: 1, t: 1538054356},{x: 1, y: 2, t: 1538054359},{x: 2, y: 3, t: 1538054370}] It should then output a probability score of it being a human. 3. You need to test your model using both human and non-human mouse movements. You can use something like [login to view URL] to simulate mouse movements. Of course you can use any bot you want for this. 4. We need at least 80% accuracy. Any questions, please ask. Thanks.