In the 1990’s a computer program called “Deep Blue” was the first machine to ever beat a world champion grandmaster in the complicated game of chess. It signaled a huge victory in the advancement of computer algorithms and their potential for the betterment of humankind.
But a recent study out of North Carolina State University is re-purposing the power of the computing algorithm to the betterment of dogs by allowing them to receive more consistent and effective training.
The authors of the study used a special harness that is connected to a cellular phone sized computer. The computer senses the body position of the dog and, upon going from a standing position to sitting, would register a tone and dispense a treat from a nearby bin.
One of the bedrock principles of dog training is the reward is linked to the correct behavior that the trainer is trying to reinforce. The entire idea behind clicker training is that the dog comes to associate the “click” with the treat, thereby allowing the trainer to pinpoint the exact moment and behavior that is reinforced.
The study showed that the algorithm was incredibly accurate at rewarding the correct behavior at 96% but was still less so than the human trainers, who clocked in at 100% accuracy. The important difference between computer and human was the in the time of response. The computer was highly consistent while the human trainers varied wildly.
This is an important breakthrough for the computer. Applying consistency to all aspects of the training will allow dogs to learn faster and more efficiently.
The authors of the study were quick to point out that this technology is in its infancy. “The study was a proof of concept, and demonstrates that this approach works….next steps include teaching dogs to perform specific behaviors on cue, and integrating computer-assisted training and human-directed training for use in various service dog applications”
This technology has some far-reaching ramifications for the training of military and service dogs. It will be interesting to follow this research as it is applied to more complicated training scenarios.