No one wants to hasten their cognitive degradation as they age. The serious reality of the long-term harm stemming from the results of sports concussions has led to a recent rethinking about the human need to protect a vitally important asset—the brain—from physical damage.
Scientists Mikhail Lebedev, Ioan Opris, and Manuel Casanova have written on and are researching the topic of brain augmentation. “Project leader Lebedev, a senior researcher at Duke University, North Carolina, said the reality of brain augmentation—where intelligence is enhanced by brain implants—will be part of everyday life by 2030, and that ‘people will have to deal with the reality of this new paradigm.’”
Technologically inclined, futuristic thinker Ray Kurzweil (1948–) articulates that human brains are very slow when compared to the speed that electronic computers can process. Despite the human brain’s inherent ability to parallel process vast amounts of information, Kurzweil believes that shortly the increasing computational speed of digital computers will far outstrip the ability of the human brain. He suggests that if scientists can learn how the brain takes chaotic and complex activities and then organizes them for understanding, this will lead to breakthroughs in computer processing that will far outpace any biological improvements that might result in increased human intelligence. Such understanding concerning the mechanisms behind the internal programming of the brain may then naturally yield improvements in artificial intelligence (AI).
Artificial Neural Networks
Progress in the area of AI has recently undergone a rapid transformation as technologists, inspired by the brain’s biological neural networks (BNNs), the foundation for human and animal thinking, have been analogously adopted into artificial neural networks (ANNs). Future developments in ANNs may lead to breakthroughs in robotic and human cognitive augmentation—providing dynamic increases in both machine and human intelligence.
ANNs involve a connected system of nodes that behave in a manner analogous to human neurons, cells that transmit nerve impulses. Neurons also can process information and make dynamic connections to other neurons. This process allows for learning. In ANNs this flow of information occurs through a complicated process that a non-linear function represents, by the sum of mathematics that enables outputted weightings to respond dynamically over time. The effect allows for reinforced learning to happen.
ANNs have made significant inroads, helping technologists in areas including machine vision, human speech recognition, and medical diagnostics. ANNs utilize the most advanced electronic components including field-programmable gate arrays (FPGAs), central processing units (CPUs), vision processing units (VPUs), digital signal processors (DSPs), AI accelerators, application-specific integrated circuits (ASICs), and system-on-chips (SoCs).
Making Possible the Future
One company, Intel®, is making possible the most amazing experiences of the future. Harnessing the latest advances in memory and programmable solutions, Intel is disrupting industries and solving global challenges by enabling all things that are smart and connected. Intel offers FPGAs, SoCs, complex programmable logic devices (CPLDs), VPUs, and complementary technologies, such as power solutions, to provide high-value solutions to customers worldwide.
FPGAs provide a flexible platform for challenging applications such as neural networks. The FPGA provides, in one sense a canvas, a sort of tabula rasa (blank slate) from which to build a foundation. The fabric inherent within the FPGA provides the intellectual property (IP) blocks and components to solve neural networking design challenges such as computing, logic, and memory resource needs.
The world of neural networks is a world full of ongoing computations. FPGA accelerators and floating-point DSP design, in combination with supporting processors, provide products with the speed, predictability, and energy efficiency to take on the ongoing big-data analytics, device virtualization, and machine learning issues inherent to ANNs. In this rapidly expanding field, reprogrammable FPGAs allow for the continual implementation of the newest algorithms and neural network topologies, ensuring high-performance computing that augments human cognitive performance. A high-performance, precision-adaptable FPGA soft processor such as the Intel Stratix 10 FPGA or the Intel Stratix® V High Bandwidth FPGA are suitable choices.
FPGAs’ sophisticated, internalized control and signal processing enable the quick and efficient movement of intensive signal processing functions. Low-power-consumption design is paramount, so that like the human brain, minimal power is consumed when neural activity is in a wait state. Offering power-consumption advantages over fixed-function graphics processing units (GPUs), FPGAs are an excellent choice. Allowing computation in parallel-processing modes accelerates performance, providing improvements in cognitive-imitating performance. The ability to incorporate vision systems, for object recognition, through sensory cameras provide a sort of electro-physical sensing that can expand over time as additional sensors are developed, contributing to the ability to take in and process information intelligently.
Today’s electronic components are allowing society to augment our intelligence. Parts, systems, and solutions that enable neural networks, which emulate and expand the capabilities of human intelligence, are opening new opportunities for both robots and humans to perceive and achieve new possibilities.