Four key concepts you need to know about AI

[51CTO.com Quick Translation] AI (artificial intelligence) is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence. Specifically speaking, every AI solution is built on four foundations. Still not clear? Check out our quick guide!

There is no doubt that artificial intelligence is sweeping the world, and endless innovative applications are being implemented in all industries and fields. As described in the movie, humans have been using artificial intelligence robots to replace doctors for decades. From experts in all walks of life to ordinary consumers, artificial intelligence is helping us diagnose and solve problems faster, such as performing delicate surgeries and playing a song with voice commands.

The general public only notices the benefits of artificial intelligence, but for professionals, there are four concepts that must be understood: classification methods, categories, machine learning, and collaborative filtering. These four pillars also represent the steps in the analysis process. Classification methods involve creating metrics for a specific problem domain (e.g., finance, network). Categories involve which data is most relevant to the problem being solved. Machine learning includes anomaly detection, clustering, deep learning, and linear regression. Collaborative filtering involves finding patterns on large data sets.

Classification method

AI requires a lot of data related to the problem being solved. The first step in creating an AI solution is to create “design intent metrics,” which are used to categorize the problem. Whether you are trying to build a system that will play a critical role in helping doctors diagnose cancer or helping IT administrators diagnose wireless problems, you need to define metrics that allow the problem to be broken down. For example, in wireless networks, the key metrics are user connection time, throughput, coverage, and roaming. In cancer diagnosis, the key metrics are white blood cell count, ethnic background, and X-ray scans.

category

Once users have categorized the problem into different areas, the next step is to segment it so that users can be directed to meaningful conclusions. For example, when the AI ​​system is dealing with critical issues, users must first write the specific problem in text form and then categorize it by time, person, and place. In wireless networks, once users know the category of the problem (such as pre- or post-connection issues), users need to start categorizing the cause of the problem: association, authentication, Dynamic Host Configuration Protocol (DHCP), or other wireless, wired, and device factors.

Machine Learning

Now that the problem has been broken down into domain-specific chunks of metadata, users can feed this information into the amazing and powerful world of machine learning. There are many machine learning algorithms and techniques, and supervised machine learning using neural networks (aka deep learning) has become one of the most popular approaches. The concept of neural networks has been around since 1949, and I built my first neural network in the 1980s. But with the innovation of computer technology and the increase in storage capacity, neural networks have been developed to solve a variety of practical problems, from image recognition to natural language processing to predict network performance. Other applications include abnormal feature discovery, time series anomalies, and event deep analysis.

Collaborative filtering

Most people experience collaborative filtering when they choose to watch a movie on Netflix or shop on Amazon and get some movie recommendations or purchase suggestions. In addition to recommendation systems, collaborative filtering is also used to solve large data sets and face recognition. This is where all the data collection and analysis turns into meaningful insights or actions. Whether it is used in game shows, doctors or network administrators, collaborative filtering is a means to provide answers with a high degree of confidence. It is like a virtual assistant that helps solve complex problems.

There is still a lot of room for development in artificial intelligence, and its impact is so profound that it will occupy a considerable share in our daily lives in the future, just like before we buy a car, we need to know what is under the hood to ensure that we buy a good product that is really suitable for us.

By Bob Friday
Original link:
http://www.infoworld.com/article/3200790/artificial-intelligence/4-key-ai-concepts-you-need-to-understand.html
Translated by Liu Nina

[Translated by 51CTO. Please indicate the original translator and source as 51CTO.com when reprinting on partner sites]