Aziana (ASX:AZK) has advised that BrainChip, which it seeks to acquire, has successfully completed its first performance milestone proving the application of its breakthrough Spiking Neuron Adaptive Processor technology.
The company is acquiring BrainChip, which has invented a neural technology that has the ability to learn autonomously, evolve, associate information and respond to stimuli, just like a biological brain.
In the stimulated race car demonstration which pitted SNAP against a sigmoid neuron network, two windows display a 2D simulated environment in which the cars drive around the track and must learn to avoid the walls.
BrainChip’s car, which uses a spiking neural network of spiking neurons with the STDP learning method, took just 25 seconds to learn the track in one pass, well ahead of the opposing car that took 15 minutes and 24 generations.
“We are delighted to receive this advice from the BrainChip team. It symbolises a major step forward in the application and commercialisation of this potentially ground-breaking technology,” managing director Neil Rinaldi said.
“In particular, the completion advice of such a milestone illustrates the rapid progress the BrainChip team is making toward commercialisation and advancement of the technology.”
This achievement has occurred ahead of schedule and is the first milestone in a series of post-acquisition milestones as the technology advanced toward commercialisation.
Consequently, Aziana will be amending its notice of meeting and shareholder approval documentation to recognise the completion of this milestone.
The other three remaining contingent milestones will be put to shareholders at the up-coming shareholders meeting to approve the transaction.
Aziana is on track to complete the acquisition of BrainChip.
Race Car Demonstration
In the demonstration, BrainChip applied its SNAP technology into a stimulated race car to demonstrate how it enabled neurons to learn rapidly and much faster than has previously been achieved by alternative solutions.
It also proves that a process can be learned and controlled using spiking neuron based learning, the same method that is used in the human brain.
BrainChip’s SNAP technology can start with “blank” neurons that contain no knowledge whatsoever, and through rapid autonomous learning produces a functional set of synaptic values that express spike timing.
The interface between the BrainChip neurons and a computer is also functioning and can configure the neural network for a specific function, in this case, learning to continually driving faster around the race-track without hitting the walls.
Both cars in the demonstration were equipped with proximity sensors that measure distances between different parts of the car and the walls.
Prior to starting the demonstration, both neural networks have no knowledge of the physical shape of the track. The learning method for each car is recorded, and their performance is compared.
In BrainChip’s car, the simulated neurons transmit information by emitting spikes at specific times the same way real neurons in the human brain transmit.
The strength of the connections between neurons, called synapses, change over time as the network learns by a method known as STDP (Synaptic Time Dependent Plasticity).
In this simulation, each time the car touches the wall, it experienced ‘pain’ from contact with the wall and learned to avoid it.
After a single pass, the car learned to drive around the track, avoiding the walls successfully and did not crash.
In contrast, the opposing network of sigmoid neurons created “generations” of solutions, and then tested each solution for how long it takes for the car to crash.
When a generation is successful it becomes a ‘parent’ to the next generation. This process is repeated until the car no longer crashes (about 24 generations in this demonstration).
BrainChip believes this demonstration is a major advance in the application of its technology and that it will continue to evolve into SnapSim, a simulator for BrainChip neurons that potential clients can use to develop applications before accessing the BrainChip platform.
Further, it believes that this Application Program Interface (“API”) will continue to be expanded and documented as a tool for programmers to access the BrainChip SNAP accelerator, further enhancing its potential as a ground-breaking commercial application.
It will now move to emulate the process by implementing a hardware emulation of the Race Car Demonstration to perform the same function using the same API and illustrate the hardware scalability of the SNAP technology.
The result from the demonstration of the SNAP technology is value accretive for Aziana shares, highlighting the rapid progress BrainChip is making towards its commercialisation and advancement.
This also marks the first in a series of post-acquisition milestones and the company has accordingly amended its documentation to recognise the accomplishment. With further series of post-acquisition milestones as the technology advanced toward commercialisation to occur we see further upside in Aziana shares after a period of settling.
BrainChip’s artificial neural learning technology has potential applications across many areas including smartphones, robotics, prostheses and driverless drones.
Aziana remains one of Proactive Investor top technology picks in 2015.
Proactive Investors Australia is the market leader in producing news, articles and research reports on ASX emerging companies with distribution in Australia, UK, North America and Hong Kong / China.