In the last article, we tried to understand what exactly is artificial intelligence. Since AI is a technology whose goal is to mimic human behavior, we can safely say that the branches of AI will comprise of those entities which make us different from machines. So let’s brief about the branches of artificial intelligence and try to correlate them with human activities!
1.Machine Learning: This is the technique that enables a computer to learn on its own by providing it with sufficient data. Much like humans, machine learning trains a system to predict an outcome using past experiences. A machine learning algorithm recognizes patterns in the given data, trains a model and predicts the outcome without having to explicitly program it for the same.
One of my professors gave a remarkable example to corroborate how machine learning is just like training a baby to face the world. A baby goes near a candle, burns his finger, and is now hurt! He couldn’t merely reason what just happened. Let’s associate this with training an algorithm. When the candle burns his finger for the second time, the baby is now cautioned and knows what might have caused the burn. This can go on for a while until the baby finally figures out that the flame from the candle is the reason why his finger burns. Now that our “model” is built, let’s test it. The next time a baby comes near a candle, it knows that the flame can harm him and totally avoids it. Safe to say that our model is successfully trained! This is exactly how machine learning takes place.
2.Neural Networks: Given the fact that it has been a buzz word for a while now, a neural network might seem like a complex term to some of us. Trust me, if you take the math out a neural network aside, it is fairly simple to understand. All you need to do is feed your model with inputs in the first layer, specify the hidden layers and the output would be your last layer. The job of hidden layers is to extract important information from the input provided, to predict the outcome. We can choose the number of hidden layers to be as many as we want but we have to be careful because it may lead to overfitting and inturn tamper the accuracy of our model.
If you’re familiar with the biology of a neuron, neural networks might be easier for you to understand. The input layer, like dendrite, is the receptor that takes the input, the neuron processes the information like the hidden layers, and the axon transfers the processed signals and acts like the output layer.
3.Robotics: What makes robotics interesting is that it is the amalgam of mechanical engineering, electrical engineering, computer science and several other scientific fields. It deals with the design, production and operation of robots, to perform the tasks that it was built to do.
Robots are the “body” of an intelligent system, it coordinates with the program and its outcomes to perform a specific function, quite similar to the skeletal and muscular system of the human body, right? It’s amazing to see how robots can be built to be so life-like, much like Sofia, the day is not far when we humans could finally have a robot for a friend!
4.Expert System: We now know how we can program a machine to learn like a human, but ever wonder how to make a machine think like a human? Well, this is where expert systems come into the picture. The expert system is an application to enable the computer to mimic the decision-making ability of humans. The three components of an expert system are user interface, inference engine and knowledge base.
Like our eyes, the user interface takes the user query and passes it on to the inference engine. The inference engine is like our brain, it has a specific sequence of rules to solve a problem and it refers to the knowledge base to provide reasoning. The knowledge base is like our memory, it is a huge repository of information obtained from experts in the domain. Hence, the success of an expert system highly depends on the accuracy of its knowledge.
5.Fuzzy Logic: We humans are highly subjected to having a dilemma, so it would only be fair if the systems that we design are trained to face such situations too. Fuzzy logic is a technique that deals with solving problems having uncertainty. Imagine looking up in the sky and seeing a few dark grey clouds on a nice sunny day. Confusing right?
Could you determine if its gonna rain or not? Could you say a ‘definite yes’ or a ‘definite no’? Here’s where fuzzy logic will help you! Unlike Boolean algebra, fuzzy logic doesn’t require the absolute values ‘True’ or ‘False’. In fact, you can have intermediate values like ‘partially true’ or ‘partially false’ when dealing with fuzzy logic. A fuzzy architecture comprises four components- rule base, fuzzification, inference engine and defuzzification. The rule base consists of a set of rules and if-then conditions provided by the experts to govern the decision making. Fuzzification is used to convert crisp inputs, (the values passed into the system for processing) into fuzzy sets. The inference system then determines the matching degree for each rule and decides which rules are to be fired accordingly. The fired inputs are then combined to form control actions. Defuzzification converts the fuzzy sets obtained from the inference engine into crisp values and then passes it on as the output.
6.Natural Language Processing: Have you ever tried to communicate with someone who didn’t speak your language, and they couldn’t understand you? Quiet challenging, right? Now imagine trying to communicate with a computer, isn’t it even more challenging? What do words and phrases mean to a computer which can only understand the language of zeroes and ones? It may not seem like an easy task to teach machines to understand our communication. Well, yes and no.
The process of making a machine read, decipher, understand and make sense out of human interaction is called natural language processing. In a nutshell, the natural language system works in the following way- A person says something to the machine, the machine records sound and turns the audio into text. The NLP system then parses the text into components, understands the context of the conversation, and the intention of the person. Based on the results, the machine determines which command should be executed. This is just how humans communicate, we listen to what the other person is saying, try to understand the meaning of their speech and then give a suitable reply in the same context. Right?
I believe it’s safe to say that artificial intelligence is even more fun when we try to correlate it with human activities. Don’t you agree? Well, that’s it for this article. In the next one, I’ll be briefing about some applications of AI, stay tuned!