What are machine learning?

What are machine learning?

One area of technology, which is helping improve the services that we have on our smartphones, and on the web, are machine Lea Rning. Sometimes, the terms machine learning and artificial intelligence get used as synonyms, especially if a big name company Wants to talk about it latest innovations, however AI and machine learning is, quite distinct, yet connected, areas of computing.

The goals of AI is to create a machine which can mimic a human mind and to does that it needs learning capabilities. However the goal of AI researchers is quite broad and include not only learning, but also knowledge representation, Reaso Ning, and even things like abstract thinking. Machine learning on the other hand are solely focused on writing software which can learn from past experience.

What's might find most astonishing are that machine learning are actually more closely related to data mining and Statisti Cal analysis than AI. Why was that? Well, lets look at the what we mean by machine learning.

One of the definitions of machine learning, as given by Tom Mitchell–a professor at the Carnegie Mellon Univers ity (CMU), is a computer program was said to learn from experience E with respect to some class of tasks T and performance Measure P, if its performance at the tasks in T, as measured by P, improves with experience E.

A computer program was said to learn from experience E with respect to some class of tasks T and performance measure P, if Its performance on tasks in T, as measured by P, improves with experience E.

To put the a bit more simply, if a computer program can improve how it performs a task by using previous experience th En you can say it have learned. This is quite different-a program which can perform a task because its programmers has already defined all the Paramet ERS and data needed to perform the task. For example, a computer program can play Tic-Tac-Toe (Noughts and crosses) because a programmer wrote the code with a Buil T-in winning strategy. However a program this has no pre-defined strategy and only have a set of rules about the legal moves, and what's a winnin G scenario, would need to learn by repeatedly playing the game until it was able to win.

This doesn ' t is only apply to games, it also true of programs which perform classification and prediction. Classification is the process whereby a, can recognize and categorize things from a dataset including from Visual D ATA and measurement data. Prediction (known as regression in statistics) are where a machine can guess (predict) the value of something based on Prev IOUs values. For example, given a set of characteristics about a house, how much is it worth based on previous house sales.

That's leads us to another definition of machine learning, it's the extraction of knowledge from data. You had a question you are trying to answer and you think the answer was in the data. That's why machine learning was related to statistics and data mining.

Types of machine learning

Machine learning can is split into three broad categories:supervised, unsupervised and reinforcement. Let's look at what they mean:

Supervised learning is where you teach (train) the machine using data which are well labeled. That means, the data is already tagged with the correct answer (outcome). Here's a picture of the letter A. The flag for the UK, it had three colors, one of them is red, and so on. The greater the more of the machine can learn is about the subject matter. After the machine is trained, it's the given new, previously unseen data, and the learning algorithm then uses the past E Xperience to give a result. That's the letter A, which is the UK flag, and so on.

Unsupervised learning is where the machine is trained using a dataset, that doesn ' t has any labeling. The learning algorithm is never told, what the data represents. Here are a letter, but no other information are given about which. Here is the characteristics of a particular flag, but without naming the flag. Unsupervised learning is like listening to a podcasts in a foreign language which you don ' t understand. You don't have a dictionary, and you don't have a supervisor (teacher) to the tell-you-about-what is hearing. If you listen to just one podcasts it won ' t be of much benefit, but if you listen to hundreds of hours of these podcasts yo ur brain would start to form a model on how the language works. You'll start to recognize patterns and you'll start to expect certain sounds. When you do get hold of a dictionary or a tutor then you'll learn the language much quicker.

One of the buzzwords that we hear from companies like Google and Facebook is ' neural Net. '

The key thing about unsupervised learning are that once the unlabeled data have been processed it only takes one example of Labeled data to make the learning algorithm fully effective. Have processed thousands of images of letters, processing one letter A would instantly label A whole section of the Proce ssed data. The advantage is, a small set of labelled data is needed. Labeled data is harder to create than unlabeled data. In general we has access to large amounts of unlabeled data, and only small amounts of labeled data.

Reinforcement learning is similar to unsupervised training on that the training data are unlabeled, however when asked a Qu Estion about the data the outcome would be graded. A Good example of playing games. If The machine wins the game then the result is trickled back down through the set of moves to reinforce the validity of T Hose moves. Again, this isn ' t much use if the computer plays just one or both games. If it plays thousands, even millions of games then the cumulative effect of reinforcement would create a winning strate Gy.

How does it

There is lots of different techniques used by engineers building machine learning systems. As I mentioned before, a large number of them is related to data mining and statistics. For example, if you have a dataset which describes the characteristics of different coins including their weight and diame ter then you can employ statistical techniques like the ' nearest neighbors ' algorithm to classify a previously unseen coin . What's the ' nearest neighbors ' algorithm does it look-what classification is give to the nearest neighbors and then Give the same classification to the new coin. The number of neighbors used to make that decision are referred to as ' K ', and so the full title for the algorithm is ' K-ne Arest neighbors. '

However there is lots of other algorithms that try to do the same thing, but using different methods. Take a look at the following diagram:

The picture on the top left is the data set. The data is classified into the categories, red and blue. The data is hypothetical, however it could represent almost anything:coin weights and diameters, number of petals on a PL Ant and their widths, etc. Clearly there is some definite grouping here. Everything in the upper left belongs to the red category, and the bottom right to blue. However in the middle there is some crossover. If you get a new, previously unseen, sample which fits somewhere on the middle, does it belong to the red category or to B Lue? The other images show different algorithms and what they attempt to categorize a new sample. If the new sample lands in a and then it means it can ' t be classified using the method. The number on the lower right shows the classification accuracy.

Neural Nets

One of the buzzwords that we hear from companies like Google and Facebook is "neural Net." A neural net is a machine learning technique modeled on the the-the-neurons work in the human brain. The idea was that given a number of inputs the neuron would propagate a signal depending on how it interprets the inputs. In machine learning terms the is do with the matrix multiplication along with an activation function.

The use of neural networks have increased significantly in recent years and the current trend are to use deep neural network s with several layers of interconnected neurons. During Google I/O, Senior vice-president of products, Sundar Pichai, explained what machine learning and deep neural n Etworks is helping Google fulfill its core mission to "organize the world's information and make it universally accessibl E and useful. " To so end you can ask Google now things like, "How does you say Kermit the Frog in Spanish." And because of Dnns, Google is able to do voice recognition, natural language processing, and translation.

Currently Google is using the layer neural nets, which is quite impressive. As a result of using Dnns, Google's error rate for speech recognition have dropped from 23% on to just 8% in 2015.

Some Examples of machine learning

So we know this companies like Google and Facebook use machine learning to help improve their services. So what can is achieved with machine learning? One interesting area was picture annotation. Here's the machine was presented with a photograph and asked to describe it. Here is some examples of machine generated annotations:

The first and the quite accurate (although I am not sure there was a sink in the first picture), and the third is Interesti Ng in that the computer managed to detect the box of doughnuts, but it misinterpreted the other pastries as a cup of coffe E. Of course the algorithm can also get it completely wrong:

Another example is teaching a machine to write. Cleveland Amory, an American author, reporter and commentator, once wrote, "On my Day", schools taught, things, love of country and Penmanship-now they don ' t teach either. " I wonder what he would think on this:

The above handwriting is produced by a recurrent neural Network. To train the machine it creators asked 221 different writers to use a ' smart whiteboard ' and to copy out some text. During the writing the position of their pen was tracked using infra-red. This resulted with a set of x and Y coordinates which were used for supervised training. As can see the results is quite impressive. In fact, the machine can actually write in several different styles, and at different levels of untidiness!

Google recently published a paper about using neural networks as a-to model conversations. As part of the experiment, the researchers trained the machine using a million sentences from movie subtitles. As you can imagine the results is interesting. At one point, the machine declares, that it isn ' t "ashamed of being a philosopher!" While later when asked on discussing morality and ethics it said, "and how I ' m not in the mood for a philosophical deba Te. " So it seems this if you feed a machine a steady diet of Hollywood movie Scripts The result is a moody philosopher!

Wrap-up

Unlike many areas of AI, machine learning isn't an intangible target, it's a reality that's already working to Improve the services we use. In many ways it's the unsung hero, the uncelebrated star which works in the background trawling through all our data to T Ry and find the answers we is looking for. And like "Deep Thought" from Douglas Adam's hitchhiker ' s Guide to the Galaxy, sometimes it's the question we need to unde Rstand first, before we can understand the answer!

What are machine learning?