Neuromorphic hardware has been around for a while but is finally gaining traction and is beginning to be applied in exciting new ways.
“It’s AI technology on a chip which uses pattern recognition technology and will be the next disruptive technology. It has the architecture similar to the neurons of the human brain,” Pierre Brunswick, CEO of NeuroMem Technologies, told TechRadar Middle East at the Ai Everything event taking place at the Dubai World Trade Centre.
Neuromorphic is critical for the fourth industrial revolution and works how human neurons work, he said. It can benefit sectors such as security, smart cities, home building, automation, autonomous driving, IoT, medical, drones and biometric.
“When you listen to music in the morning, you don’t need to open a computer to know the singer. You do pattern recognition. Pattern recognition is what we do in our neurons. The chip does exactly what our neurons are matching,” he said.
Four pillars of AI
According to Brunswick, AI has four pillars - deep learning, reinforcement learning, edge computing and neuromorphic hardware.
Reinforcement learning is the opposite of deep learning, which means learning from the mistakes and by itself. For the last five to six years, He said that a lot of spending is focused only on deep learning.
“There is no way AI can only be [about] deep learning. The winner will be the one who will be first on all the four pillars. AI is everywhere. If the journey for developing AI into the business is long, you have to embark early.” he said.
Will robots have human intelligence?
In the last 25 years, NeuroMem has sold its IP technology to quite a few companies, including Intel.
“We are an evangelist and help companies develop proof of concepts faster. Our technology is so critical as it uses low power, non-stop learning and anomaly detection in milliseconds. We believe that if we come closer to the human neurons, we have come closer to the truth,” said Brunswick.
However, he believes that hardware cannot replicate the human brain and robots cannot have human intelligence- but that could change in the future.
“We can easily make up to one billion neurons with neuromorphic chips but the human brain has 10 billion neurons. Each neuron in the brain simulates 1,000s of synapses, the electricity going from the brain to give action. We can come close to what the neurons are doing but we are far away from copying the human brain. The goal is not to replace humans but to make our lives easier and better,” he said.
It's all about ethics
In the present, companies are collecting data, understanding it and treating it much better than any humans according to Brunswick.
“We are creating petabytes of data and if we don’t put in place a really good AI platform to extract what needs to be extracted from the data, it would be a nightmare. On top of that, we need to put ethics on whatever we do and it is critical.
“The biggest bacteria or the destroyer of the planet are humans. This is why governments need to put a protocol to protect humans.” He added that ethics are very important and will be the centre of everything.
Computing power is essential to AI; he said and added that neuromorphic hardware enables new applications at the edge computing.