Add support for other models: different data shapes etc. #3

New issue

Open

opened 2026-03-10 14:11:17 +00:00 by leon-adsk · 0 comments

leon-adsk commented

2026-03-10 14:11:17 +00:00

Owner

Two methods:

Built-in Modules: Here we simply provide a way to specify which features should be extracted and in what order the feature vector should be constructed using our built-in Go feature extraction code. Here we should add more modules.
External Python FE: When the Go reverse proxy starts we also start a lightweight Python FastAPI. This reads scripts from a feature extraction server and uses the functions defined there to build an internal API that the Go program can call to extract features both for bulk processing, live detection and threshold finding.

It seems to me that option 2 is the most flexible and desirable so we will proceed for now. We will determine whether the built in methods for feature extraction should remain at all (and be extended in the way described in 1) or if the Python system should fully replace this one.

We will use ZeroMQ for the communication between the Python and Go processes.

Two methods: 1. Built-in Modules: Here we simply provide a way to specify which features should be extracted and in what order the feature vector should be constructed using our built-in Go feature extraction code. Here we should add more modules. 2. External Python FE: When the Go reverse proxy starts we also start a lightweight Python FastAPI. This reads scripts from a feature extraction server and uses the functions defined there to build an internal API that the Go program can call to extract features both for bulk processing, live detection and threshold finding. It seems to me that option 2 is the most flexible and desirable so we will proceed for now. We will determine whether the built in methods for feature extraction should remain at all (and be extended in the way described in 1) or if the Python system should fully replace this one. We will use [ZeroMQ](https://zeromq.org/) for the communication between the Python and Go processes.

leon-adsk added the

enhancement

label

2026-03-10 14:11:17 +00:00

leon-adsk self-assigned this

2026-03-10 14:11:17 +00:00

leon-adsk added this to the http-anomaly-detection project

2026-03-10 14:11:17 +00:00

leon-adsk added this to the Release v1.0 milestone

2026-03-10 14:13:39 +00:00

leon-adsk changed title from ~~Add support for other models: bring your own onnx~~ to Add support for other models: different data shapes etc.

2026-03-17 15:55:41 +00:00

leon-adsk added a new dependency

2026-03-28 12:50:30 +00:00

#4 Model Chaining Support

leon-adsk added a new dependency

2026-03-28 17:06:34 +00:00

#13 Advanced Threshold Algorithms

leon-adsk referenced this issue

2026-03-28 17:09:20 +00:00

Additional Threshold Algorithms #12

leon-adsk started working

2026-03-31 17:13:09 +00:00