My team works on deploying models on the edge (android mobile devices). The data, model, code, everything resides on the client device. Is there any way to protect your model from being probed into by the client? The data and predictions can be unencrypted. Please let me know your thoughts on this and any resources you can point me to. Thanks!