The Marvel of the human brain: Unravelling neural networks

The human brain, an intricate and powerful organ, serves as the epicentre of our thoughts, emotions, and actions. This marvel of nature comprises nearly 100 billion neurons, interconnected in an unimaginably complex network that enables everything from basic survival functions to the most advanced cognitive processes. Though it constitutes only about 2% of our body weight (approximately 1.2 to 1.5 kg), it consumes around 25% of the body's energy. The brain represents the core identity of a person.

While transplants of organs such as lungs, heart, kidneys, and liver, or even facial reconstructions, are now possible and sometimes logical, brain transplants do not hold the same meaning, as they would essentially involve transferring one's identity. The brain is the most efficient information processing machine ever to exist, defining one's mental architecture and emotional and cognitive abilities.

Therefore, studying and understanding the human brain is of paramount importance which must encompass both its biological structure and its functional and processing models to better understand its workings. One of the revolutionary outcomes of this is the development of artificial neural networks (ANN), which enable artificial deep learning (ADL) to solve previously unsolvable problems.

Neurons, the fundamental units of the brain, communicate through electrical and chemical signals. Each neuron primarily consists of a cell body (SOMA), dendrites (signal receptor channel), and an axon (transmitter channel). Dendrites receive incoming signals through neurotransmitters at synaptic regions, where synaptic clefts, the gapped connections, facilitate chemical reactions. These signals are processed in the cell body.

When the accumulated signals reach a certain threshold, the SOMA releases a signal that travels through the axon to other neurons via synapses. This process occurs continuously at speeds ranging from about 3.6 to 432 kilometres per hour. This communication forms the basis of neural networks, intricate webs of interconnected neurons that process and transmit information.

In modern computer processors, electrical signals travel nearly at the speed of light, around 1.08 billion kilometres per hour. This means that processor signal speeds are roughly 300 million times faster than the slowest neuron signals and approximately 2.5 million times faster than the fastest neuron signals.

Despite this stark contrast in speed, modern processors lag behind the human brain's neural architecture in several key areas, including parallel processing, learning, adaptability, energy efficiency, creativity, problem-solving, generalisation capability, contextual understanding, sensory integration, and emotional and social intelligence.

Modern computer processing architecture is not designed to support the parallel processing and adaptability inherent to the human brain. Despite its complex processing capabilities, the human brain operates on about 20 watts of power, functioning silently in a dark region submerged in spinal fluid. In contrast, modern supercomputers and artificial intelligence systems require thousands of watts to operate. The brain's energy efficiency makes it highly effective for sustained and diverse cognitive tasks.

Approximately 100 billion neurons, each connected to thousands of other neurons, enable massive parallel processing, allowing humans to perform multiple tasks simultaneously, such as perceiving the environment, processing complex sensory information, and making real-time decisions. The human brain can learn from experience and adapt to new information through the restructuring and remodelling of neural connections, a phenomenon known as neuroplasticity.

The human brain excels in creative thinking and problem-solving, often finding innovative solutions that current artificial systems cannot achieve. Neural networks in the brain are not static; they exhibit remarkable plasticity, allowing the brain to adapt and reorganise itself in response to new experiences, learning, and injury.

This plasticity underpins critical functions such as memory, learning, and recovery from brain damage. The ability to understand context, make intuitive leaps, and generate novel ideas are unique strengths of the human brain. Artificial systems, on the other hand, often require specific programming for each task and struggle with transferring knowledge from one domain to another.

The brain integrates information from multiple sensory inputs; sight, sound, touch, taste, smell, etc.; to create a coherent perception of the environment. It is astonishing how the brain can sense heat, cold, pain, touch, pressure, and more from a tiny surface of the human body resembling a complex sensory module.

Considering these points and supporting them with scientific research, it can be argued that the human brain's neural network is the best data processing system in the universe in terms of versatility, efficiency, and capability.

Recent research has significantly advanced our understanding of neural networks. Sophisticated imaging techniques, such as functional magnetic resonance imaging (fMRI) and positron emission tomography (PET), have enabled scientists to observe brain activity in real-time and map neural connections with greater precision. One ground-breaking discovery involves the concept of Neural Coding, which explores how information is represented in the brain.

Researchers have found that specific patterns of neural activity correlate with particular thoughts, perceptions, and actions. This has led to advances in brain-computer interfaces (BCIs), allowing direct communication between the brain and external devices by ensuring hundred percent dissemination of information. Moreover, BCIs have shown promise in restoring mobility to individuals with paralysis and aiding in the treatment of neurological disorders.

Additionally, recent studies are focusing on designing artificial hippocampus to mimic biological navigation memory and processing. Complementing this research there are specifically designed transistors optimised for artificial intelligence (AI) applications.

These are often referred to as AI transistors or neuromorphic transistors and include examples such as memristors, spintronic transistors, carbon nanotube field-effect transistors (CNTFETs), phase-change memory (PCM) transistors, and organic electrochemical transistors (OECTs). These advanced transistors are part of ongoing research and development aimed at creating more efficient and powerful AI hardware.

They serve as non-volatile memory that can emulate the synaptic functions of neurons, enabling neuromorphic computing to build circuits that mimic the brain's neural networks. OECTs, used in bioelectronics, can interface with biological systems and are being researched for use in AI to create bio-hybrid systems that integrate electronic and biological components. Though, these technologies are not yet widely available in commercial products, they are being actively explored in academic and industrial research labs.

The principles of neural networks have also inspired the field of AI. Artificial neural networks (ANNs), designed to mimic the structure and function of the human brain, have revolutionised machine learning and AI. These systems consist of layers of interconnected artificial neurons that process information and learn from data, enabling tasks such as image recognition, natural language processing, and predictive analytics.

Deep learning, a subset of machine learning, utilises multi-layered ANNs to model complex patterns and make sophisticated predictions. Recent advancements in deep learning have led to significant breakthroughs, including the development of AI models capable of outperforming humans in specific tasks, such as playing complex games and diagnosing medical conditions from imaging data.

Understanding neural networks holds profound implications for medicine. Insights into how neural circuits function and malfunction can lead to improved treatments for neurological and psychiatric disorders.

For instance, research into neural networks has contributed to the development of more effective therapies for conditions such as epilepsy, depression, and schizophrenia. AI-driven diagnostic tools, powered by neural network algorithms, can analyse vast amounts of medical data with unprecedented accuracy, leading to earlier detection of diseases and personalised treatment plans.

The human brain, with its intricate neural networks, remains one of the most fascinating and least understood frontiers of science. Recent advances in neuroscience and artificial intelligence are unlocking the mysteries of these networks, offering new possibilities for medicine, technology, and our understanding of the human mind.

As research continues to evolve, the potential to enhance human health, cognition, and interaction with technology grows ever more promising, heralding a new era of discovery and innovation.

The integration of neuroscience and AI-based learning has the potential to transform healthcare, finance, transportation, education, food management, industry, politics, government, defense, and all other sectors of human life to a superior level. It is astonishing that a simple equation of a straight line, modelled by a single artificial neuron within a complex artificial network, can perform a variety of tasks far better than humans.

It is evident that those at the forefront of developing and utilising AI are leading the world. However, the core control of AI should not be independent, as AI-driven robotic behaviour and thinking (Robotanity) must not overshadow humanity.

Dr M Akhtaruzzaman is an Assistant Professor in the Department of Computer Science and Engineering, Military Institute of Science and Technology (MIST), Bangladesh. He is also associated with Quantum Robotics and Automation Research Group (QRARG) and DAZDREAM Technologies as a consultant and research scientist. Email: akhter900@gmail.com.

Disclaimer: The views and opinions expressed in this article are those of the author and do not necessarily reflect the opinions and views of The Business Standard.