Perceptual Detector Learning

1. Parameters

Transition Distance 4.0

Dist < T: Excite. Dist > T: Inhibit.

Good Continuation (5x5) 2.0

Adaptive Noise (Max) 1.0

High noise if detector is scattered

Category Unlearning 0.1

Active category erases OFF pixels

Conscience Influence 20

Boosts under-used detectors

LR (Non-Winners) 0.01

Passive Decay 0.05

Lateral Inhibition 0.2

Presets

Controls

Ready

Click a pattern in preview to train on just that one.

2. Input Pattern

Click to Toggle. Drag to Paint/Erase.

Current Category Label:

A B C None

Dataset Preview

3. Learned Detectors

Weights: Green=1, Black=0

4. Category Units

Weights: Detectors → Categories

Weights -1 (Black) to +1 (Green)