I’m fairly new to this subject and I am working on a project that deals with detecting anomalies in time-series data. I want to use TensorFlow so that I could potentially deploy the model onto a mobile device. I’m having a difficult time finding relevant material and examples of anomaly detection algorithms implemented in TensorFlow.
Some algorithms I’m looking into are clustering algorithms for classifying windowed samples and Holt-Winters for streaming data.
Any example would help me tremendously!
If you’re interested in a Hierarchical Temporal Memory implementation in Tensorflow, there is one that I know of here: https://github.com/calclavia/htm-tensorflow (disclaimer: I’ve not tried it and it’s been over a year since the last commit). It looks fairly basic and it might need a bit of work to do anomaly detection.
Since you’re specifically targeting mobile, it could be worth checking out this example mobile app: https://github.com/numenta/numenta-apps/tree/master/taurus-mobile
Hope this helps!
Thanks For Your Reply, It’s very helpful for me