What it is: Machine learning is a way for software to adapt and learn without outside human intervention.
In the early days of artificial intelligence, scientists made bold promises that computers would one day think. Yet decades passed and artificial intelligence failed to impress anyone. The big problem was that computer scientists were trying to create artificial intelligence by programming a computer to be as knowledgeable as a human right from the start.
That goal is impossible because even humans don’t know how to model knowledge effectively. One of the biggest flops in artificial intelligence was expert systems. The idea behind an expert system was that you took the knowledge of a human expert and distilled it into a series of if-then rules.
Unfortunately, expert systems had two huge flaws. First, translating someone’s knowledge about a domain such as heart surgery or engine repair was extremely difficult. That often meant expert systems held incomplete information, making them far from perfect from the human expert they were modeled after.
A second problem was that expert systems were static. Once you programmed it, it could never learn. That meant you had to constantly update the expert system. Since modeling knowledge accurately was a problem in the first place, constantly updating the expert system was equally problematic. That meant expert systems often were incomplete and obsolete, which made them largely useless. That’s why you rarely hear of expert systems today.
Machine learning is different. Instead of requiring constant outside updating by humans, machine learning allows the program to adapt to changing conditions with no input from humans whatsoever. Not only is this faster, but also more accurate because now the computer can update its own knowledge as rapidly as conditions change. Many hedge funds use machine learning algorithms to trade stocks. Since the stock market constantly changes, these algorithms constantly change to adopt to the latest market conditions.
If you’re interested in machine learning, you may want to learn more. Before buying a book about machine learning, try a free ebook instead. Most of these ebooks are written for a scientific audience so if you’re willing to wade through complex topics that are often not explained in layman’s terms, you can teach yourself more about machine learning.
The future is machine learning. The smarter algorithms get on their own, the more versatile and flexible they’ll be without requiring constant human attention. Apple, google, Microsoft, and many other big companies are already pursing machine learning so it’s something you can expect to use on a daily basis, often without even realizing you’re using a smart computer.
Every time you use Siri, you’re using a natural language recognition program along with a machine learning algorithm that gradually gets smarter about responding to common queries. You may never need to know how machine learning actually works, but you may want to know more about machine learning just to understand its benefits and limitations.
One thing for sure, machine learning can never become self-aware and take over the world. That’s what politicians are for.
To find a list of free machine learning ebooks, click here.