Эксперимент: Cross-Channel Scattering Front End (#1405)

{ "worked": true, "confidence": 7, "verdict": "Built a differentiable fixed-Morlet Cross-Channel Scattering front end with local edge pairing, complex conjugate products, differentiable magnitude, and temporal averaging. The numerical checks confirmed |z_m z_n*|=|z_m||z_n|, phase invariance, and exact agreement with an independent implementation to floating-point precision. On the scarce-label synthetic amplitude-coupling task, SNST achieved 0.5406 +/- 0.0181 accuracy versus 0.5104 +/- 0.0053 for the raw CNN across three seeds, a promising but modest signal; this is not evidence of a real-EEG win.", "metrics": { "baseline": "Raw temporal CNN: held-out accuracy [0.5156, 0.5125, 0.5031], mean 0.5104, std 0.0053", "idea": "SNST front end plus classifier: held-out accuracy [0.5156, 0.5484, 0.5578], mean 0.5406, std 0.0181; +0.0302 absolute mean accuracy" }, "how_to_run": "/home/maxwelhelp/main/bin/python3 experiment.py", "files": [ "experiment.py" ], "limitations": "Tested only on a small synthetic four-channel amplitude-coupling task, not BCI Competition IV-2a or real EEG. Only three seeds were used; no channelwise-scattering baseline, ablations, learned wavelets, strict FLOP/parameter matching, or subject-wise evaluation was performed. CUDA convolution failed on the available runtime and the script correctly fell back to CPU." }