Researchers Predict that Machine Learning is expected to outdoes human experts in next few decades
After adjustment by experts, it can leave human beings far behind in any problem that can be calculated purely by logical analysis. The speed of the machine will be faster and faster, the learning ability will become stronger and stronger, and the data will be more and more.
We will see countless opportunities and products that solve problems, save lives, and generate tremendous business and user value. In the future, automated trading can achieve higher return on investment and risk ratio. Automatic diagnosis + genetic sequencing will achieve personalized precision medicine. The recommendation engine will be able to recommend the products you are most likely to buy, the dishes you want to eat, and the people you want to know.
According to the report, machine learning will have a huge impact on all aspects of people’s lives. This is a reality, and it will happen more quickly than most of us realize. According to the news network on June 4th, the US media said that if someone is still alive in 2100, then they may see machine learning as a decisive advance in the 21st century when reviewing the past. Then they will let a robot write a blog post about it.
The Google AutoML system has recently produced a series of machine learning code that is even more efficient than the researchers themselves. Obviously, this is another blow to the “human superiority theory”, because the robot “student” has become the master of “self-replication”.
In the absence of high level programming talent for machine learning the AutoML was developed as a solution.The team proposed a machine learning programming that could make self-learning code that would run a thousands of simulations to figure out which parts of the code could be enhanced and proceed until the point when the objective was accomplished.
Although it sounds a little scary, AutoML does far better in programming machine learning systems than the researchers who created it. In an image recognition task, it achieved a record 82% accuracy.
Even in some complex artificial intelligence tasks, its self-created code is superior to human programmers. It can mark multiple points in the image with an accuracy rate of 42%; in contrast, humans build only 39% of the software.
Of course, it doesn’t mean “Skynet” or the creepy “digital ghost”, because we are not on the edge of the “self-perceived machine”, but we add another to the technical potential of artificial intelligence.
