Machine learning - you can hear the drumbeat in the distance!
Since I was about 10 or 11 years old I have had a personal passion and interest in "machine learning", which I've always personally looked at "what it will take to make computers able to do what people can do, and more. I studied Cognitive Science at Carnegie Mellon University and have maintained an ongoing interest in the developments of this field. A few announcements in the past month (see links below) have caught my eye and made me want to share some thoughts.
Most people fundamentally misunderstand the dynamics in the field of machine learning and artificial intelligence ("AI"). They look at the current severe limitations of software, computers and machines and state in a million ways how, even with their amazing features, they still don't exhibit real "intelligence". While this is true, it fails to recognize that machine learning is going to fundamentally change what is possible. Over simplified, machine learning is a process whereby software can be "taught" through experience in same way that humans can. No one programs in all the rules - the software uses methods to watch and learn from patterns of information.
The primary reason that machine learning methods haven't completely changed the everyday world yet (though they've already done more than you likely realize) is that the speed of the underlying computer hardware has been too slow to accomplish much. Machine learning requires very fast, highly parallel systems to work well for solving complicated problems.
Enter Moore's Law, the Internet, and the progression of time...
What's happening now is that computers that are widely available are reaching speeds that are beginning to make it possible to go after some really, really interesting problems. Couple that with the increasing speed of the Internet and computer networks, and you also unleash opportunities for massive parallel processing.
So what does this mean?
It means that we're going to start seeing a ton of interesting applications where computers will exhibit signs of "intelligence". These will start to percolate their way into our everyday lives, and people are going to be blown away, excited, threatened, and a myriad other reactions by what will become possible.
Here are a few examples that are emerging now, many of which you may not think of as "machine learning":
- StumbleUpon learns your browsing patterns and helps you discover other sites that you are likely to be interested in.
- My6Sense learns about your reading behaviors and helps you find just the news and information that is most important to you.
- Picasa suddenly supports facial recognition in an announcement yesterday.
- Vitamin D launches object recognition for video.
These are just the harbinger of what's to come, and it will all represent a wave of change that will likely be even more profound than what the Internet has brought to us.
Exciting times ahead!
- 5 months ago
- • 0 comments (Add)
- • Share
