If you still think it’s all bitcoins and blockchain out there, here’s a mini-challenge for you. Try to browse the daily newspaper on the internet without seeing a single reference to machine learning. It is virtually impossible! It’s the concept of machine learning and its possibilities that is now capturing the public’s imagination.
With all this chatter going on, let’s get a clearer understanding of the machine learning phenomenon. Some of the questions that come to mind are: What does ‘Machine Learning’ actually mean? Can machines learn? If so, what are they learning? Who is teaching which machines to do what? Whatever these machines are being taught, is it a task, a skill, or simply a database of information?
The Textbook Definition
Machine Learning (ML) is defined as the application of artificial intelligence (AI) that enables systems with the ability to learn and improve from experience. This learning and improvement take place automatically. All of this happens without the machine being specifically programmed to do so. Learning, in this context, means the progressive improvement in the performance of any one task. This learning is often based on data or a series of observations that is completely free of human assistance or interference. We provide the examples and the computer should be able to come up with a program to execute a task automatically.
Researchers have been searching for a way to build a computer that keeps improving and expanding its intelligence. Arthur Samuel first pioneered ‘machine learning’ as a term in 1959 while developing a program at IBM. Samuel was successfully able to train the IBM 701 computer to play a game of checkers better than he could. What contemporary scientists understand is that the application of machine learning is a two-step process. The first step involves creating computers that can learn and the second step involves teaching them how to learn.
Two Peas in a Pod: Machine Learning & Computer Vision
Smart machines put artificial learning and machine learning to work for the benefit of people like you and me. Machine Learning and Computer Vision belong to the larger Computer Science discipline. They are complementary but can function equally well on their own.
Computer Vision involves getting the computer to detect or ‘sense’ a movement, a gain, a loss or a void. A textbook definition of computer vision describes it as the ability “to make useful decisions about real physical objects and scenes based on sensed images”. Machine Learning, on the other hand, consists of getting the computer to ‘recognize’ a pattern so it can understand what is going on. Next, the pattern is matched to a pre-existing library or knowledgebase. It is at this point that machine learning takes the lead. Before any pattern matching can take place, someone must construct the knowledgebase with samples and train the computer system on how to use it.
Facial recognition software, such as the algorithms used by Facebook and Snapchat to apply filters, is a commonly-found application of this partnership. Other joint applications of machine learning-computer vision include biometrics via fingerprint-matching, motion-sensing for surveillance in public spaces, gesture recognition, vehicle detection for traffic monitoring purposes, and automated lip-reading for speech recognition.
Machine Learning- All Day, Every Day
Humans use machines, simple ones and complex ones, to accomplish their different objectives. In fact, we want each and every task to be done as quickly and as efficiently as possible. We should just apply Machine Learning to meet these twin aims. Machine Learning provides us with the option to delegate all types of mental labor such as classification, recognition, research and even personalization, like Netflix recommendations.
Machine learning surrounds us every single day – language translation services like Google Translate, the auto-filtering and categorization of our email inboxes, as well as the growing presence of digital assistants like Amazon’s Alexa. Apple products use Siri, a voice-recognition system, to understand requests and to carry out them out while maintaining a conversation. Social media platforms, like Instagram, employ machine learning algorithms to translate the meaning behind emojis used in captions. Once the platform has understood the contextual meaning, it can then suggest relevant hashtags and emojis to the user.
Let’s dig a little deeper into the inbox scenario. Each time the user marks an email with a label, the email service actively ‘learns’ that classification preference and then suggests the same label for similar emails in the future.
You probably use the voice feature on the smartphone to do your most urgent tasks. Let’s suppose you want to get the weekly grocery shopping done while stuck in a traffic jam. Your smartphone is using voice-to-text technology to convert voice into text and then run a search for the query. This entire process mimics a human brain. It uses neural networks to make each voice search faster and more accurate than the last.
How Deep Is Your Learning?
Deep Learning, also known as the Deep Neural Network, is a strand of Artificial Intelligence concerned with mimicking the approach humans use to gain knowledge. Deep Learning involves stacking algorithms in a hierarchical system that gets increasingly more abstract and complex. Fresh input helps clarify to the computer what is actually going on. This clarification usually results in the creation of a brand-new featureset that helps the machine identify one concept from another.
A distinct advantage in Deep Learning is Unsupervised Learning, so a program can build a featureset without any direct supervision. All the received information takes form in a brand-new featureset which eventually results in the program building a predictive model. Deep Learning programming has the capacity to convert large amounts of unstructured data into labeled, predictive models.
Why is Deep Learning is all the rage now? Simply because immense amounts of processing power and sample data were not easily accessible until recent times. The shift occurred when cloud computing and big data became more commonplace.
Machine Learning and Mankind: A New Phase
No matter how you slice it, machine learning continues to gain influence in our lives. From how we commute to how we communicate, smart machines are taking up a greater role in our day-to-day existence. At this point in human civilization, we really can not afford to ignore our growing dependency on tools and systems. We all should be familiar with technical topics, like machine learning, artificial intelligence, or augmented reality, solely because it is all around us. Knowledge brings with it the power to navigate this quickly changing technological terrain to our collective advantage.