Data Science in Plain Terms
What is Data Science?
Data Science studies methods of data analysis and ways to get valuable information (knowledge) from it. It is closely related to such fields of study as Machine Learning and Cognitive Science (study of thinking), as well as technologies of working with Big Data.
Since the start of massive spread of technology, humans have generated huge amounts of data. The scale of data is so big humans are not able to visualize and process it. Information about our calls and movement, things we do online, shopping preferences, anthropogenic changes of landscapes, climatic processes and many other things are all examples of Big Data. If processed correctly, Big Data can be a source of highly valuable information.
It used to be the case that computers gained new abilities by means of programming. That is, humans created working algorithms for machines, which the machines could understand and reach predictable goals. The tendency has changed, as this approach is becoming out-of-date.
Working with Big Data requires a different approach, and Machine Learning came in handy. Under this approach, humans provide machines with input data and step aside. This means that the results of such algorithm in action are not determined by humans. Humans specify the ways in which a machine can learn, but the machine then learns on its own, reaches conclusions, and analyses information. It resembles the way humans learn. Machine Learning is not only about Artificial Intelligence. This field of study includes algorithms for gene research and study of evolution, as well as much simpler tasks, dealing with cluster analysis, for example.
As to Cognitive Science, it is a multidisciplinary study dealing with mechanisms of cognition and thinking. Results of Cognitive Science research become the basis for developing various approaches to Artificial Intelligence.
What do neural networks have to do with it?
Neural networks are self-learning networks (i.e. Machine Learning technology) created in the image and likeness of a human brain. Such networks use Big Data as the material they learn on. In other words, neural networks are the most progressive product of Data Science which humanity currently has.
Neural networks try to replicate some structural principles of neural networks of a human brain, and that is why the name. Neural networks are not able to replicate human brain in full, as they are not powerful enough yet. According to Maksym Orlovskyi, the mentor at Cloud Business City, the first virtual business center in the cloud that develops Data Science projects, it will take some 30 to 50 years until neural networks can match a human brain.
There are two types of Artificial Intelligence, namely: general intelligence (similar to human intelligence) and specialized forms of intelligence. So, neural networks can be of general and specialized types.
Specialized types are capable of solving particular tasks, often being far more efficient than a human is. It is this type of neural networks that is rapidly becoming popular and is seen valuable on the current stage of technological development.
Neural networks is not a completely new idea. The technology, construction approaches, and key algorithms for learning were developed back in 1950s and 1960s. Nevertheless, only in recent years have the key factors come together, allowing Artificial Intelligence to make a leap forward in quality in terms of computing power, available sets of Big Data, and thoroughly worked-through frameworks.
Neural networks can be applied almost anywhere now. In judicial science, for instance, they can be used to find precedent cases (which is especially important in judicial system of the USA). In the realm of financial technologies – to analyze transactions, support loyalty programs, track customer loyalty, etc. In logistics – to predict necessity for specific goods. In medical treatment, by means of processing vast amounts of data neural networks can find unexpected factors, which influence patient’s health, and can diagnose even the most difficult medical conditions.
Neural networks are applied in creative activities as well. But the well-known image filters are only a by-product, emerging out of the wish of developers to find out the principles of neural network learning algorithm. It has become popular with mass audience. Much less is known, though, about successful application of neural networks in translation practices, data recognition and processing. Meanwhile, these technologies can be used in projects by adding corresponding functions using API – leading companies in technological industry provide this opportunity.
Neural networks are already capable of painting in the manner of famous artists, such as the “New Rembrandt” project, realized with the support of Microsoft. But the question of when neural networks will be able to create pieces of art on their own is still to be answered. This may become a reality when the power of AI exceeds the abilities of human intellect.
New professions and old professions
Any new sphere of activity gives rise to new professions. Data Scientist, a specialist on working with data, and Machine Learning specialist are the most wanted specialists in future. They are not programmers or developers. They are brilliant mathematicians with a superb ability for analysis and persistence, because chances of finding an ideal formula for machine learning at the first take are close to zero. They have to be able to find the right algorithm among all of the existing ones to suit the needs of a project in the best possible way, and when something goes wrong, they have to be able to understand what it is.
Data scientists must be able to understand what form a set of data should be given for a computer to be able to process it. Their main task is to provide this data. Machine learning specialists help data scientists by choosing architecture and learning algorithms to work with a set of data.
Giants of IT industry have given free access to their frameworks, which render writing thousands of lines of code unnecessary. These frameworks allow anyone with basic knowledge of programming languages to create neural networks. For instance, Microsoft Azure cloud contains all the instruments necessary for working with Artificial Intelligence, and one does not need to buy powerful hardware and expensive software in order to start using Artificial Intelligence in their start-up. All the instruments are available as “product as a service”, and one just has to activate them in cloud.
Why Data Science is important
Data Science and artificial intelligence technologies make it possible to learn more about what people prefer (by gathering and analyzing data), become closer to them by means of creating more personalized interfaces (e.g.: selecting offers on the basis of what was interesting for a user and sending personalized mail), etc.
In IT industry the opportunity of working with data is akin to a huge quantum jump as it is impossible to imagine any new start-up without employing this technology. Failing to use Data Science would be like using horses for transportation in the high of the era of cars. Moreover, “IT start-up” as a term presupposes innovation.
Automatizing and introducing new ways of personalization allows increasing business marginality. And if you don’t make use of these technologies, competitors who do will outrun you and you’ll lose your business.
So, when will AI kill us?
So far, we can say it is not happening anytime soon. Theoretically speaking, only general artificial intelligence can possess such power in case it supersedes human intelligence. But such artificial intelligence has not been created yet.
Want to learn more about Data Science, Machine Learning or Neural Networks? Contact us now.