|
|
|||||
|
|
|||||
|
|
Jet Propulsion Laboratory California Institute of Technology The 10 Rules by NASA |
|
Q - A
Reliable Machine Learning Platform for no integration efforts A fully managed and reliable Machine
Learning Platform from our partner that contains all you need to build easily
high quality ML applications. With this platform you can deliver ML models into
production faster and easier than ever before. This platform facilitates the ability to
build, train, deploy, monitor and maintain ML and AI models in production at
a scale. ·
Fastest time to market ·
Zero dependency on infrastructure teams ·
Best TCO compared with competitors ·
Standardized ML workflows |
|
|
ASPRET-MS™ Pattern Recognition Framework ASPRET-MS™ is the complete modular system
and development framework including ASPRET™, NHTMR™, PRTVA™ technologies,
Desert™ features extraction DL algorithm and Shalom™, Laser™, Sharp™,
Rocket™, Mythos™ pattern recognition algorithms. The OmegaPhiBuilder™ tool is
required for the preparation of training and testing datasets with ASPRET™
technology. We can supply these functionalities only within EU/DARPA funded
research projects. |
|
|
Mythos™ Mythos™ (virtual SIMD Parallelism on
Von Neumann machines) technology has been developed for Machine Learning
solutions that must run on Rad-Hard CPU like BAE SYSTEMS RAD750™ and any Single-Core CPU operating with
safety-critical RTOS such as Green-Hills® Integrity™
(although this RTOS supports AMP and SMP multiprocessing). Lockheed Martin® is using
Green Hills Software's INTEGRITY®-178B RTOS safety-critical
and security-critical software for the Lockheed Martin® F-35 Joint Strike
Fighter (JSF).
|
|
|
Laser™ Laser™ neural network
has been developed for Pattern Recognition with anti-spoofing technology and
can be associated with Deep Learning neural networks. This neural network can
work as stand-alone ML solution with conventional features extraction
algorithms. |
|
|
Sharp™ Sharp™ neural
network has been developed for Machine Learning applications that require
explainable inference. This library enables ML inference with recovery of the
original training examples. This is
the neural model with the highest explainability and is extremely fast in
learning and inference. |
|
|
The C source code
of our algorithms is optimized for C2VHDL conversion. This feature ensures
portability of our algorithms on Radiation Tolerant/Hardened FPGA. The C2VHDL
conversion process seems simple but generates very inefficient VHDL codes and
the subsequent synthesis phase generates large bit-streams that cannot be
loaded on most FPGAs. The process of optimizing the VHDL code is often too
complex and time-consuming. Our C code is optimized for C2VHDL conversion and
eliminates the need to modify the VHDL code generated by the conversion.
|
|
|
©2024_Luca_Marchese_All_Rights_Reserved Aerospace_&_Defence_Machine_Learning_Company VAT:_IT0267070992 NATO_CAGE_CODE:_AK845 Email:_luca.marchese@synaptics.org |
|
Contacts_and_Social_Media
|
