So, what is the first thing that comes to mind when you hear the term Artificial Intelligence or AI? — most probably think of an interactive robot, a chat bot or any kind of device which is smart enough to answer your intriguing questions and perform the intended action without much fuss, in just the way you command.
Some may imagine a conflicted robotic system who decides to destroy you or the world around you as you know it. 🙂
In short, you may be expecting something out of a science fiction movie, which is ‘intelligent’ enough to understand and interpret everything just like any other human being, if not better and may have a mind of its own.
Since this term is so widely used and often incorrectly many people find this terminology quite confusing so let’s start by simplifying it with a good definition. Here is one is from techopedia.com.
“Artificial intelligence (AI) is an area of computer science that emphasizes the creation of intelligent machines that work and react like humans.”
The science behind AI is actually quite broad and you can think of similarly to a tree, where AI is a trunk with multiple branches, sprouting off. (Image 1)
Image 1: The many branches of AIIn this blog series I will explore many of the various branches of AI and their application in our world today. We’ll start at the top with the one in particular that has gained a lot of interest lately Machine Learning(ML).
Although the term Machine Learning has become a buzzword in recent years, it is a term which dates back to 1959.
In simple language, it can be defined as the ability of a machine to perform certain tasks based upon its previous experience, with the efficiency of the machine to carry out these tasks and improving with every experience. In technical terms, the experience which the machine gains is the training done on the model, through which it evaluates its own losses for better performance the next time a dataset is provided to it.
How about some practical examples.
Do you remember when you first learned to ride a bicycle? I bet you didn’t just hop on and go. You probably started with training wheels, then lots and lots of practice. Probably after taking off those training wheels and falling down time and time again, you gradually learned to ride that bike. If you can relate to this example, then you can relate to a type of Machine Learning algorithm called Reinforcement Learning! Simply put, you can think of this as a reward based learning approach similar to how through your trials and tribulations you eventually learned to ride a bike.
As another example, try to think about a time when a friend or coworker had such poor handwriting you could never read what they had written. Eventually though with enough exposure you learn to read and understand their writing most, if not all, the time.
How does this happen? Lets explore.
In the beginning you might have had to frequently ask your friend about what he or she had written. There might have even been times when you tried to guess some illegible text but got it wrong(misclassified). Slowly your ‘guesswork’ started to improve based upon the feedback you received from your friend in the form of corrections for the misclassifications. In this example you just learned another Machine Learning approach, known as Supervised Learning.
Putting it in technical terms, the misclassifications is the loss function which needs to be reduced and the guesswork which you did was prediction or classification to be more precise. This method of finding the class(type) of an object based upon its parameters (in this case the length of the character, its thickness, etc.) is what is known as Supervised Learning. This method can also be applied for predicting certain numerical values based upon certain parameters, known as Regression. Thus, you now have knowledge of two forms of Supervised ML Algorithms — Classification(for categorical values) and Regression(for numerical/continuous values).
Okay, enough of my supervising you with these algorithms. Let’s move on to some Unsupervised Learning.
Let us consider an example where you are given multiple types of pictures of various kinds of animals yet you are not told how many types of animals are there. What would you do? You will surely try to cluster similar looking pictures and separate them from the other clusters based upon how the pictures look, things like the color of the eyes, number of legs, body hair distribution, etc. This is exactly what a machine does as well, it tries to form clusters on the basis of certain features, which it thinks are important. The items in each cluster are similar while they are dissimilar from the items in the other clusters. Whoa! Maybe this ML stuff is not so complex after all, if you’re following along so far you now understand yet another ML algorithm — Unsupervised Learning.
One last thing worth a mention here is Deep Learning. Deep learning is another application of Machine Learning and is a bit harder to understand. Essentially deep learning involves a lot of complex computations which works in a similar manner to that in which our brains function. It is based upon Artificial Neural Networks(ANN) or Convolutional Neural Networks(CNN) and as the names suggests, ANNs & CNNs are networks of nodes similar to the neural network in our brains. Applying these techniques allows for the powerful computations required to attain higher levels of accuracy than were previously possible.
A practical use for this is in autonomous vehicles. Consider the task of needing to quickly identify and determine various objects, for example the difference between a stop sign and a pedestrian, important no? The computations for this happen many times a second and the results of poor accuracy in the outcome could be dire.
Given the potential complexity of Deep Learning we will further explore it separately in an upcoming blog.
While this is meant as an introduction to AI you can see that the topic does not have to be overly complicated or difficult to understand. It is after all more an abstraction in which an electric powered device (the machine) tries to imitate the thought process of a human being and match up to their intelligence.
I hope you enjoyed reading this and you found this information was a helpful starting point in your journey to understanding what AI really is. Also I’m curious as to your impressions on the real world examples and their helpfulness in relating machine processings similarities to our own human learning processes. It is our goal at QueryAI to reduce complexity and enable understanding to anyone who wants to learn!
If you like this content or have suggestions for other topics you’d like us to cover please let us know, we’d love to hear from you. Please leave a comment in the form below!
Follow us on linkedin and visit www.query.ai to subscribe to our updates.
Thanks for stopping by!
