/Artificial%20Intelligence%20IoT%20Blog.jpg)
The world-wide buzz about The The Internet of Things (IoT) has opened discussions for a myriad of futuristic ideas, but the conversation is shifting. Now, the tech world is trying to figure out just how to make these innovative ideas actually work and it looks like artificial intelligence may be the answer.
One of the biggest obstacles with devices connected to the IoT is the question of how to sift through and analyze the vast sea of data these devices will collect in order to make that information valuable.
Improving the speed and accuracy of analyzing massive amounts of data is vital for The IoT. If analytical options do not advance, the outcome could range from simply annoying, think appliances in your home that don’t connect like they promised, to life threatening, like incidents that result in piles of self-driving cars.
According to an article from WIRED, “The only way to keep up with this IoT-generated data and gain the hidden insight it holds is with machine learning.” Machine learning, as defined by Wikipedia, is “a subset of artificial intelligence in the field of computer science that often uses statistical techniques to give computers the ability to "learn" with data, without being explicitly programmed. ”
DeepMind, a subsidiary of Google, is currently the world leader in artificial intelligence research. At the International Conference on Machine Learning in Stockholm, Sweden this past week, they presented on their new paper “Measuring Abstract Reasoning in Neural Networks” which is based around the question: can machines learn to reason abstractly?
The researchers at DeepMind defined abstract reasoning as a capacity to identify patterns and solve conceptual problems. To measure this, the team was able to create a test that essentially evaluates the IQ of individual AI systems through tests fairly similar to ones used to test human intelligence.
Of their findings, DeepMind said, “Altogether, we introduce and explore ways to both measure and induce stronger abstract reasoning in neural networks. Our freely-available dataset should motivate further progress in this direction.”
The ability for future AI systems and machine learning to abstractly reason will be invaluable. AI detecting the patterns, correlations and abnormalities found in the overabundance of information the IoT can improve almost every aspect of our daily lives.
