A summary of Chapter 1 and 4 of “The Master Algorithm” by Pedro Domingo
Intelligence? What is intelligence? How can you really define? How you proof that a person is Intelligent and how can you prove that a machine is intelligent? can we say that the way a machine thinks is the way a human being can think? Can one process information better than the other? These questions can be seen as common or uncommon to society in this day and age but what makes a human mind unique compared to the machine mind or vice versa. In this blog I will talk about how the human mind works and how machine mine works. Hopefully by the end of this you get a good feel or understanding on how close we are to matching human intelligence. I will be summarizing what was said in Pedro Domingo’s book, “The Master Algorithm”, touching chapters one and four where he talks about what an algorithm is and how the human brain works. I will be using these chapters interchangeably because I personally find that the human body to be a unique and complicated machine but yet it is not a machine. This is not about how one is better the other but to show or somewhat reveal similar traits between human intelligence and Artificial Intelligence.
An algorithm can be defined as a set processes that dictate activities. Most common functionality is acquiring an input and creating an output and/or outcome. Pedro Domingo defined an algorithm as a sequence of instructions telling a computer what to do. A computer can contain billions of transistors that can be used by such algorithms. They can be combined together to form information as a collection of on and off states. An algorithm can be as complicated or as simple as the creator of the algorithm wants it to be. This can be defined in its time and space complexity Can depend on the creator. obviously the less of each complexity the better the alchemy. The same case applies to human complexity which you say is far vaster than the complexly of a machine.
Can we consider a human brain as organic processor containing a multitude of algorithm? I say of course we can. Consider daily human activity such as eating, reading writing, doing an exam, coming up with a new groundbreaking company idea. The resulting action and/or output is a result of processing inputs like stomach pain caused by hunger or noticing that you company stock price is dropping. We grow up learning how to live in the world that we are in and continue this education down the line. A good example would be an athlete of any sport. An athlete would train their body and mind in order to win a game or assist a team of team of other athletes in winning. This training can consider the pros and cons of each action taken on the field.
The same thing applies to a machine. A data scientist can create an algorithm that can be used to determine improvisation to be applied to a web application or a mobile application and achieve better marketplace results.
So how does the human brain work? The human brain Is divided into 3 main parts; the brain stem, the cerebellum and the cerebrum. The cerebrum, which is the largest section of the brain, is divided into two hemispheres each of which has four lobes; the frontal, temporal, parietal and occipital lobes. The frontal lobe deals with the cognitive and behavioral aspects of the human brain. The temporal lobe is responsible for hearing and understanding language. The parietal lobe is responsible for the sense of touch, pain and temperature. Finally, the occipital lobe is primarily for vision.
Each neuron is like a tiny
tree, with a prodigious number of roots—the dendrites—and a slender,
sinuous trunk—the axon
When it comes to the frontal lobe and its cognitive purposes, this part of the brain has an elaborate series of functions. Problem solving, writing, speaking language, intelligence, concentration and alertness are just a few of the many things this part of the brain can accomplish. In the book, Domingo mentioned the firing of neurons. He referenced the book Principals of Psychology by William James, stating the principle of learning by association similar to Hebb’s rule Where neurons replaced by brain processes and a firing efficiency by propagation of excitement (Domingo). The basic process of learning is association, correlation; when two things occur at the same time, or one closely follows the other, then link your representation of these two things. In brains the link occurs at synapses, where the axon of one neuron contacts a dendrite of another (Westbrook). As Donald Hebb said, “Neurons that fire together wire together.” (Domingo).
“Neurons that fire together wire together.”
Donald Hebb
Machines, however, have a similar concept but instead of neurons a machine’s brain consists of transistor. According to Domingo’s differences between symbolist learning and connectionist learning, we understand that computers learn and/or process information in a sequential order while the brain does so in parallel. Computers require a multitude of step to accomplish a task where transistors can switch between States at a rate of a billion times per second. However, the brain will process billions of neurons at the same time but at a rate of a thousand times per second. Domingo stated that the number of transitions in a computer is catching up to the number of neurons in the brain, but the brain has more connections than a computer (Domingo). The computer can match the brain’s connectivity but speeding up the process transitioning through one of its few connections. In term of resources the amount of power the brain uses is significantly smaller compared to the amount that a computer uses.
A neuron has been described to be like a tiny tree with a prodigious number of roots called the dendrites and a slender chunk known as the axon. Brain is stated to be a collection or a forest of such trees where the tree branches make connections called synapses to the roots of other neurons. These connections run electrical signals that create actions called potentials. Domingo gave a good example of toe wiggling.Domingo gave a good example of toe wiggling.
When you wiggle your toe, a series of electric discharges, called action potentials, runs all the way down your spinal cord and leg until it reaches your toe muscles and tell them to move
Pedro Domingo
The brain is a synaptic result the electrical sparks that formulate such action potentials. According to Domingo, synapses do grow when post dramatic in neurons fired soon after presynaptic. Neurons have different concentrations of ions inside and out creating a voltage across the membrane. He further mentions that when presynaptic neurons are fired tiny sacs release neurotransmitter molecules into the synaptic cleft causing channels in the postsynaptic neuron’s membrane to open. This lets in potassium and sodium ions and changing voltage across the membrane as a result.
Domingo introduced the term Perceptron a variable of weights to the connections between neurons. In a perceptron a positive weight represents an excitatory connection while a negative weight represents an inhibitory connection. Its outputs vary between 1 and 0 where 1 shows that the weight of the input is above the threshold while the latter shows that it is below. This results some information being left out but if what was left out is important, it can add it later. The higher an input’s weight, the stronger the corresponding synapse (Domingo). The perceptron fires when the resulting inputs have a weight of one and the number of these inputs passes more than half of the desired threshold. Domingo gave an example using a grandmother cell, which defines a cell that fires whenever you see your grandmother. You can train the perceptron to remember your grandmother. This could be through an image that shows the grandmother’s physical features. If the test show that the cell doesn’t fire when an Image of the grand mother is shown, then more training is required otherwise no training is needed.
In a brain when a cell receives an input and it is already depolarized from other inputs it will strengthen that input. An active cell represents the presence of some event (object, movement). Strengthening the connection between two concurrently active cells forms a physical record of the correlation between them. That’s the basics of how learning works in the brain, how information is stored. There are many other activity-dependent processes in neurons influencing excitability, but they are more about adaptation than correlation (Westbrook).
A typical neuron would occasionally spike in the absence overstimulation, it would spike more frequently as the stimulation builds up and saturates at the fastest spiking rate beyond which increased stimulation has no effect. in comparison to a logic gate a neuron is more like a voltage to frequency converter (Domingo). The curve of frequency as a function of voltage is represented as an S curve shown below. The curve shows how the output increases slowly in correspondence with the input and seemed almost constant but then the output goes faster and then slower to a point where it seems constant again. Domingo states that both computers and the brain are filled with S curves. He further describes it as the shape of phase transitions of all kinds this includes the probability of an electron flipping its spins as a function of an applied field, the magnetization of iron, the writing of a bit of memory to a hard disk and ice melting to name but a few.
This curve, which looks like an elongated S, is variously known as the
logistic, sigmoid, or S curve.
FAU CAP6635 Artificial Inteligence 
One thought on “Human Intellect vs Machine Intellect”