CogniMem CM1K
Neural Network Chip for High Performance Pattern Recognition
CM1K Neural Network Features
CM1K Neural Network Features
- 1024 neurons
- automatic model generation in learning mode
- 256 byte neuron memory size
- up to 32766 categories with 16-bit distance metric
- adjustable influence fields
- up to 128 sub-networks
CM1K Neural Network Classifiers
CM1K Neural Network Classifiers
- Radial Basis Function
- K-Nearest Neighbors
CM1K Neural Network Metrics
CM1K Neural Network Distance Metrics
-
L1 Norm
-
Lsup Norm
Development Cycle - General
Train/Execute Development Cycle
Development Cycle - Imaging
Image Recognition Application
Development Cycle - Audio
Audio Recognition Application
CogniMem CM-E2K Expansion Board
- Two CM1K Neural Network chips giving 2048 neurons
- Stackable to four expansion boards:
- 8192 neurons
CM.1Kit
CM-1Kit Development Kit Features
- high performance pattern recognition made easy and cheap
- easy connection to a processor and sensor(s)
- 1 CM-1K Neural Network Chip
- standard header for breadboarding
CogniMem Evaluation Boards
- CM-EB evaluation board
CM-IRB image evaluation board
- USB interface
- Documentation
- SDK
- Diagnostics
- Video trainer (CM-IRB)
- Optional enclosure and tripod