Neuromorphic computing is rivaling Quantum computing



Original Source Here

Neuromorphic computing is rivaling Quantum computing

Neuromorphic computing seeks to build hardware and software systems that are inspired by the brain. It is an emerging field of computer science and engineering that explores the design of hardware and software systems that mimic the computational abilities of the brain. The term “neuromorphic” comes from the Greek words for “nerve” and “form.

The goal of neuromorphic computing is to create systems that can learn and adapt in ways similar to the brain. These systems could potentially be used to solve problems that are difficult or impossible for current computers, such as recognizing patterns in data or making predictions based on incomplete information. This new way of architecture promises to create systems that are more efficient, more flexible, and more scalable than current computers. The brain is an example of a highly efficient and scalable system. It can learn new tasks quickly and adapt to changes in its environment. Neuromorphic computing seeks to replicate these properties in artificial systems.

There are a number of challenges associated with building neuromorphic systems. One challenge is understanding how the brain works. Another challenge is designing hardware and software that can effectively mimic the workings of the brain.

Current computers are based on a von Neumann architecture, which is not well suited for neuromorphic computing. In the von Neumann architecture, each task is performed by a separate processor. This can lead to bottlenecking and inefficiency when multiple tasks are being processed simultaneously.

In contrast, the brain uses a parallel processing architecture, where multiple tasks are performed simultaneously by different parts of the brain. This architecture is more efficient and allows the brain to handle more complex tasks than a von Neumann computer.

One approach to neuromorphic computing is to use analog circuits instead of digital circuits. Analog circuits are better suited for parallel processing than digital circuits. However, analog circuits are more difficult to design and build than digital circuits.

Another approach to neuromorphic computing is to use custom-designed chips that incorporate both digital and analog circuitry. These chips are designed to perform specific tasks such as pattern recognition or image processing. Custom-designed chips offer the advantage of flexibility but come with a higher cost than general-purpose processors.

Industry analysis for neuromorphic computing

Gartner has placed neuromorphic computing in its “innovation trigger” stage and included it on two of its latest Hype Cycles. While neuromorphic architectures have the potential to be viable within the next five years, it may be a decade before they reach the productivity stage.

Several companies are working on neuromorphic computing technologies, including IBM, Google, Qualcomm, and Intel. IBM’s TrueNorth chip is perhaps the best-known example of a commercial neuromorphic system. TrueNorth is designed to simulate the workings of a neural network, with each chip containing four thousand processors that are interconnected like neurons in the brain.

Google has also been working on artificial neural networks that are capable of learning like the brain. In 2015, they revealed their work on a chip called the Tensor Processing Unit (TPU), which is specifically designed for machine learning applications. Google has said that TPUs can provide up to 180 teraflops (trillion floating point operations per second) of performance while consuming just 40 watts of power.

Qualcomm has been working on neuromorphic technologies for several years and has demoed a number of prototype chipsets. In 2016, they announced their Zeroth platform, which uses machine learning algorithms to enable mobile devices to understand and respond to their environment.

Intel acquired Nervana Systems, a startup working on deep learning chipsets, in 2016. Nervana’s technology is based on what are called “neural processors,” which are designed to accelerate deep learning workloads. Intel plans to use Nervana’s technology in a variety of applications, including data center servers and autonomous vehicles.

Other startups working on neuromorphic computing include:

SynSense claims that its “ultra-low power and ultra-low latency” processors might power autonomous robots, wearable healthcare systems, and always-on mobile devices.

AnotherBrain is experimenting with neuromorphic computing in quality assurance in automotive and appliance manufacturing.

ABR markets Nengo is planning to release a Time Series Processor chip in 2023 that will allow for low-cost processing using large AI models for time-series data.

Brainchip might be the first commercial neuromorphic IP with its Akida processor.

GrAI Matter Labs touts world’s first “sparsity-native AI System-on-Chip.”

Elon Musk’s Neuralink

Not exactly neuromorphic computing, but we need to touch upon Elon Musk’s initiatives here too. Tesla is working on autonomous robots and Neuralink on human brain interface. These companies could also collaborate to research on neuromorphic computing in the future.

Neuralink is a brain-computer interface (BCI) company founded by SpaceX and Tesla CEO Elon Musk, based out of San Francisco, California. The company is developing implantable brain–machine interfaces (BMIs) to treat neurological disorders, and ultimately to link humans with artificial intelligence.

Musk has said that the first application for Neuralink will be to help people with quadriplegia, and that eventually the technology could be used to enhance human cognition. He has also said that he believes Neuralink could help solve some of the most pressing problems facing humanity, such as climate change and existential risks from artificial intelligence.

In August 2020, Neuralink announced that it had achieved a significant milestone in its development of BMIs, having successfully implanted them in rats and monkeys, allowing them to control computer cursors and robotic arms with their thoughts. The company plans to begin clinical trials in humans within the next two years. If successful, Neuralink’s technology could have profound implications for humanity, potentially allowing us to merge with artificial intelligence or greatly enhance our cognitive abilities. However, the technology also raises ethical concerns about potential abuse and misuse. It will be important to carefully consider these issues as the technology develops.

AI/ML

Trending AI/ML Article Identified & Digested via Granola by Ramsey Elbasheer; a Machine-Driven RSS Bot

%d bloggers like this: