At North Carolina University, Dr. David Roberts is working on making smart devices that learn through the lens of dog training. As an esteemed computer scientist, he is the only one in the country working in the field of machine learning and is using this technique to teach machines how to learn through experience.
Making the Connection Between Dog Training and Machine Learning
Machine learning is the concept of creating algorithms to enable computerized devices to learn through experience. This is seen already in a simple form when Google remembers your browsing history to suggest search results. This is similar to reinforcement learning of the 1950’s, which was used primarily by psychologists.
That initial work was based on a system of reward and punishment which obviously will not work on a cell phone, but the concept is similar. Dogs are trained primarily in this reward and punishment system. Take a look at pet containment systems. If the dog tries to cross the line, he is punished by a small zap of electricity to his neck. These K9 fence systems work fantastic on the canine mind. The question is, how do you zap your phone?
If you could take the types of feedback that a person gives with pet training, Dr. Roberts believes you could take the guesswork out of making more intuitive computers.
Since around 40% of American households have a dog, techniques for training them are fairly common. This will make it easier on the user to work with their device so that it is “trained” properly.
Some of the Possible Problems
The reward and punishment scheme in dog training can lead to further issues if the dog misunderstands what they are being rewarded for. Untraining a dog is even harder than training one. The same is true with programming. If you teach your phone to accidentally dial mom when you say wife, it is very hard to undo that as it has become a learned behavior by an electronic device.
The goal is not only to train computers and phones to be smarter, but also other household devices. Imagine if your coffee maker learned when you wanted it on and off by your daily actions, or if your thermostat automatically adjusted the temperature when it sensed rain outside.
That is a brave new world indeed, and one we are not as far from as you may think. As this technology begins to take off, expect for your devices to start operating all on their own.