Any ideas or tips for my GPS tracking RNNs project?

I am finishing course 5 of the specialization. All the courses are excellent, and the knowledge that I have acquired is really exciting. Thanks to the teachers and classmates; The tips in the community were very helpful.

I am already starting a project where I will apply the acquired knowledge. The project consists of detecting GPS spoofing in augmented reality games.

I think the most similar exercise that I could base myself on is “emojify”, since the idea is to feed an RNN with a sequence of GPS coordinates and the output would be one: “1” there is spoofing or “0” there is no spoofing.

But to feed the RNN I need to encode each GPS coordinate in the style of word embeddings.

Any idea how to perform “word embeddings” with GPS coordinates?

I will be very grateful for any other advice.

Thank you very much.

Positions are floating point values for latitude and longitude. Why do want to use word embeddings?

Thank you very much teacher for your answer.

I would like the RNN to learn that these coordinates can only be streets, parks, or others that can be passed by a person on foot or by vehicle.

So I guess I need coordinates enriched with this semantic information, encoding everything before input.