Artificial Intelligence: In Math I Trust
Artificial Intelligence: In Math I Trust
Learn about some of the real applications of artificial intelligence, machine learning, and deep learning, and see why the fear of AI is unwarranted.
Join the DZone community and get the full member experience.Join For Free
The most visionary programmers today dream of what a robot could do, just like their counterparts in 1976 dreamed of what personal computers could do. Read more on MistyRobotics.com and enter to win your own Misty.
Artificial intelligence has been shaping our world since the 1970s, or even before. There were three big moments of investment going into artificial intelligence:
- Neural networks and statistical machine learning algorithms, which are inspired by the general information processing strategy of the brain. Later in this article, we will talk more about them.
- Expert systems that became some of the first truly successful forms of artificial intelligence (AI) software. They are knowledge-based systems composed of two sub-systems: the knowledge base and the inference engine. The knowledge base represents facts about the world. The inference engine is an automated reasoning system that evaluates the current state of the knowledge-base, applies relevant rules, and asserts new knowledge into the knowledge base. The main idea is that intelligent systems derive their power from the knowledge they possess rather than from the specific formalisms and inference schemes they use.
- Genetic algorithms, support vector machines, clustering, and supervised learning models with associated learning algorithms that analyze data used for classification and regression analysis.
It started in 1950 when a handful of pioneers from the nascent field of computer science started asking whether computers could be made to "think."
Nowadays, most of the AI shown on TV and media is harmful and dangerous for our population (i.e. mad robots trying to destroy the earth or terminators in pursuit of taking over the earth). Still, far from that futuristic scenario, I am going to discuss some of the real applications that AI has and what is at the core of this new machine intelligence.
What Is Artificial intelligence?
AI is nothing but intelligence thrust into machines. It is inspired by neural networks, but actually, they are a very complex mathematical interpolation. Neural networks are loosely inspired by how the biological brain works. However, neuroscientists have always tried to avoid this term due to the confusion that it may create. AI is about learning through experience by changing connection strengths, defining how strongly neurons influence each other. It goes through three phases: learning, execution, and self-correction. It basically inserts the “experience” factor so that the computer can learn and improve every time a certain action is made.
To do that, AI uses a process called machine learning that gives computers the ability to learn without being explicitly programmed, as well as deep learning, which is a subset of architectures in the field of artificial neural networks.
Let’s start from inside in and make our way to the big picture.
Currently, deep learning is the most widespread AI field, as it brings machine learning and biological-type thinking closer. Deep learning is deep structured learning or hierarchical learning. It is part of a broader family of machine learning methods based on learning data representations as opposed to task-specific algorithms. It mimics the biological brain's neural networks in order to make a certain number of patterns using big data and a lot of computer power. Deep learning architectures such as deep neural networks, deep belief networks, and recurrent neural networks have been applied to fields including computer vision, speech recognition, natural language processing, audio recognition, social network filtering, machine translation, and bioinformatics. It is becoming more and more demanded by companies that want to be more efficient and want to innovate faster than their competitors.
Historically, computer methods had been very bad at recognizing patterns or relationships; with neural networks, it became much more simple. It breaks the complex relations down into a sense of simpler ones.
When we work with little amounts of data, methods like SVM (support vector machine) could be a good option. However, with the current tendency of IoT, everything is going digital and companies are starting to work all the time with more and more sets of data. To be able to manage those amounts of data, we have to implement deep learning into our strategy.
The picture would look like this:
Machine learning is used by the biggest companies to announce their latest innovations. It is a software trained with an algorithm that allows it to learn from past information and from humans' experiences that generates insights from the data they have encountered and apply it to future decisions. Machine learning is well-known as a predictive analytics field. Machine learning allows researchers, data scientists, engineers, and analysts to produce reliable, repeatable decisions and results and to uncover hidden insights through learning from historical relationships and trends in the data.
For example, a computer program is said to learn from experience E with respect to tasks T and performs measure P if its performance at tasks in T, as measured by P, improves with experience E. If a computer program can improve the performance of a certain class based on past experience, then you can say it has learned.
We usually categorize machine learning into three different levels: supervised learning, unsupervised learning, and reinforcement learning.
Supervised learning: Starting from the analysis of a known training set, the algorithm develops a function to make predictions about the output values.
Unsupervised learning: It explores the data and can draw inferences from datasets to describe hidden structures from unlabeled data.
Reinforcement learning: This method is about interacting with its environment by actioning parts and finding errors or rewards. It allows the system to find the best behavior within its environment to maximize its performance.
Many philosophers considered life to be mechanical.
There has always been a need in the history of humanity to understand the human-thinking method and how biological thinking has been, since the beginning, the most powerful machine.
Brands and companies have always fought to better understand, own, and manage all those amounts of information that we are using all the time. Right now, most of the instruments and tools that you use on a daily basis are based on digital functions. That results in huge quantities of data that tell companies everything about us — what we like, what we say, for how long we do something, what we're willing and not willing to change, etc. And AI helps those companies manage and control that information, increasing the efficiency of the processes.
The big five — Apple, Google, Microsoft, Facebook, and Amazon — are implementing their business market and products using deep learning neural networks.
In 2015 Google introduced a tool to improve their speech recognition, Amazon launched their AI product named Alexa, and Apple launched Quicktype. Coming back to Google, in 2016, the 30% of all the computational power for Google data scientists to analyze was used for LSTM.
LSTMs (long short-term memory networks) are neural networks used by deep learning algorithms. Because of them, Facebook was able to improve more than 4.5 billion translations a day.
Artificial Intelligence: Case Study
A few weeks ago, I attended a conference and a guy who was giving a talk mentioned that he was a big fan of Amazon. He loved everything about it — from its business model, technology, and how they're growing and to the strategy the follow to tackle and eliminate every single competitor from the market by giving their customers better service. If we look at their logo, we will see an arrow starting from the first A and pointing to the Z. From A to Z, they have everything — they own the market. Everything you can think of, you can find on Amazon.
He detailed one special product that Amazon launched with pure artificial intelligence: Alexa. It is another reference to the entire capacity of knowledge (Alexandria library). Alexa is the modern and updated version of Apple’s Siri. It can control your fridge and tell you that you are running out of beers, and with only one click on the app, Amazon Go brings beers right to your home in less than an hour.
Another example: Alexa has voice recognition, and that means that it can educate your kids. When your kids ask Alexa to buy donuts, it can give a suggestion computed before by their parents; for example: “Why don’t you go outside to run instead?” or “What if we get apples instead?”
All of that is done by using AI. I believe that there is more to come after the recent partnership of Amazon with Microsoft. They put all those neuron-type systems into algorithms that add the “experience factor” to the equation.
Some people believe that despite the positive effects that AI will bring to the society, it will mostly destroy it, eliminating some of the human aspects of the production process. The most catastrophic point of view even claims that it will see us as less intelligence pieces of life and will eliminate us. My point of view is rather near to those who are excited. Just like with the Industrial Revolution, AI will transform society, making it evolve and changing some of the current jobs.
Published at DZone with permission of David Ayza Enero . See the original article here.
Opinions expressed by DZone contributors are their own.