|
|
|
|
Safety_Critical_Machine_Learning |
|
|
Registers-Optimized neuro-Chip
Knowledge Emulation Technology (ROCKET™) is a software
simulator of a neural chip based on RBF (Radial Basis Function) architecture
and RCE (Restricted Coulomb Energy) learning algorithm. Unlike a software
"emulator" which can simply reproduce the functionality at the API
level, the software "simulator" reproduces all the registers of the
chip and performs bitwise operations on the registers in an extremely
efficient way. The simulator was written in C language using all the potentialities
that the language offers (inside our safety critical subset) to speed up
execution on Von Neumann machines. The simulation uses a proprietary
algorithm that speeds up the "broadcasting" operation on a Von
Neumann machine. This simulator cannot parallelize operations on vector
prototypes as the chip does: the hardware chip is based on a scalable SIMD
(Single Instruction Multiple Data) architecture. This simulator can be
considered the most efficient software implementation of an RBF-like neural
network architecture with RCE learning algorithm (classifier). The ROCKET™
simulator has been updated to be compliant with all the evolutions of the
chip from ZISC36® to ZISC78® to NM500® and ANM5500®. In the software simulation the number of
neurons is unlimited for all the versions. This neuro-Chip software
simulation has been successfully applied in the biomedical sector within an
ECG recognizer and in robotics within an anomaly detector for prognostic
maintenance.
|
|
|
|
General Synaptics Aerospace_and_Defence_Machine_Learning_Company VAT NUMBER:_IT02670700992 REA NUMBER: GE-503104 Email:_luca.marchese@synaptics.org |
|
|
|
Copyright© 2026 General Synaptics |
