Is it possible to create an artificial brain model or replica of the human mind?
For the past week, Google Research hosted an online session. The session discusses particularly about the conceptual understanding of deep learning of the human brain. The online session featured lectures from award-winning computer scientists and neuroscientists. It focused on how recent discoveries of deep learning and neuroscience can aid in the development of creating better artificial intelligence systems.
While all of the discussions and presentations were worth attending, one of them particularly stood out.
Christo Papadrimitriou’s lecture about the human brain
A session conducted by Mr. Christos Papadimitriou, a professor of computer science at the University of Columbia, tackled word representations in the brain.
During his lecture, Papadimitriou explained how our knowledge of information-processing systems keeps progressing. He also explained how it may contribute to the development of algorithms. These algorithms may be more efficient when it comes to understanding conversations.
Christos spoke and passed on knowledgeable facts through the digital screen. Aside from that, he provided a simple and efficient model to display. The model shows how various parts of the brain communicate with one another to solve certain problems. He said during his session, “What’s occurring now is probably one of the world’s greatest scenery.”
The brain converts structured information into airwaves. These airwaves travel through various mediums and reach the listener’s ears. During this process, the airwaves will transform and change into structured information within the brain of the listener’s ears.
Cognitive and neuroscience researchers are searching for ways to understand how neural activity in the brain translates to logic, reasoning, and other activities.
If ever scientists succeed in articulating the workings of the brain in terms of mathematical models, then they will pave the way to artificial intelligence that will emulate the human mind.
Numerous investigations were made mostly focused on activities at the single-cell level of neurons. Not until recently, scientists believed that single neurons linked to single thoughts. The most well-known example is the “grandmother cell” idea. It says that: when you encounter your grandma, a single neuron in your brain spikes. However, there are other researchers debunking this notion.
It demonstrates huge groups of neurons that relate to each thought. There also may be overlaps between neurons that link to different concepts. These are groups of neurons, which are also known as “assemblies.” Papadimitriou defines these neurons as a “highly linked and stable set of neurons that represents a word, an idea, etc.”
A mathematical simulation of the brain
Christos offers a mathematical model of the brain termed “interacting recurrent nets” in order to fully comprehend the role of assemblies.
Within each region, there is a recursion. It means the neurons interact with one another; each of these sectors are linked to several different areas. These inter-area relations have the chance to stimulate or inhibit.
This mathematical model of the brain provides randomness, plasticity, and inhibition. Neurons in each brain region link randomly in the areas, which are referred to as randomness. In addition to that, distinct areas have random connections with one another. Next in line is plasticity. It allows the connection between neurons and regions to change as a result from experience and training. And last but not the least, inhibition. Inhibition means that only a small number of neurons can stimulate at any given time.