According to me, ‘How’ is the wrong question? “Will it?” is most appropriate one. Truth be told, neurologists and neuro-researchers across the world, are still puzzled by activity of human brain. All we know till now, is there are 86 billion neurons in human mind, and they have a complex neural network, that responds to external and internal stimuli in splits of second. The mystery behind the processing is still not be decoded. Unless, it is, our limited knowledge on neuroscience is rather helping to build machine learning systems, rather than the opposite.
How to read human brain?
Human brain has been put under experiments using MRI and EEG while the brain is working or thinking deeply about something. For instance, studying a designer’s brain while at work, provides large amount of electrical impulses, which, if decoded, can answer the ‘why’, ‘how’ and ‘what’ of brain functioning. But, its yet to be decoded.
Aspects of human brain functioning
The two main brain functioning that interests people in the field of Machine Learning is – learning or acting from experience or memory, and quick response from attention. In the first case, human brain cognizes a subset of inputs immediately in a given scenario, and acts out on the best possible reward. The catch is, out of plethora of inputs, it zeroes down the best set of inputs which will have the best possible outcome. In the second case, human brain learns from scenarios and stores it in it memory zone. And when faced with similar situations, it draws out actions from the memory it learned.
Machine learning into play
Interpreting these basic two functionalities of human brain, into a machine, has taken these many years. What we have now, is near to perfection in mimicking these properties. The key element for any machine learning program is data. A huge set of data. Algorithms need to be taught first, like a baby who crawls around to learn things and then walks and learn more things, and continues to grow. To learn, it needs data. We are fortunate enough to live in a techno savvy online world, where there is data surplus. Using data, algorithms are taught to recognize different things. They can relate and find patterns in data sets, and come to a conclusive output. If you feed an algorithm at least 1000 images of an apple, next time when you connect the algorithm with a camera and it can focus on an apple, it actually can point out it’s an apple. But this is just a basic step. To make computers think like human, they need to build the neural system that a human brain has, which is the experimentation point. Here, deep learning is the next best approach of building a neural network similar to human brain network.
What is neural network?
Neural networks are set of algorithms based on different mathematical functions like linear regression, Bayesian model, to teach machines using reinforcement techniques. Neural networks have many layers, and the data sets are filtered and analysed to understand a certain pattern or recognize a specific correlation, so that the machine can conclude something based on the initial inputs. Machines becomes goal driven. Once these machines are put online, or are fed constant various data, these interconnected networks of algorithm, find a certain solution and pave their own way to create an output and act on it. Once this action is done and is successful, they retain this learning as data, and utilize these data sets to grow. Hence deep learning leads in creation of more self-evolving algorithm models. Eventually, algorithms learn too much, to understand what useful data to use among a heap of data, and lead to a conclusion in the fastest possible way.
Machine Learning mimicking the brain
It has been found out in certain cases, that different algorithms when given a group of task, and can talk with each other to consult or barter the task responsibilities. In many cases, the same model has been utilized in physical world, so that two different electronic devices can communicate and exchange data signals to achieve a task together. Hence, IoT devices have great potential to finally create a seamless ubiquitous experience among brands to ultimately lead a perfect intuitive consumer journey for the consumers. Autonomous cars work on these techniques, where they are initially fed with data sets to understand how to recognise signs, and obstacles, and drive through them. Later on, when they start driving using initial learning, they keep on learning through their driving experiences, and share the experiences in form of data that can be decoded by other cars as well.
The day is no longer far, where machine learning can actually decode the human brain science revealing the mystery behind its functioning.