“Hey Siri, how’s my hair?”
Your smartphone may soon be able to give you an honest answer, thanks to a new machine learning algorithm designed by U of T Engineering researchers Parham Aarabi (ECE) and Wenzhi Guo (ECE MASc 1T5).
The team designed an algorithm that learns directly from human instructions, rather than an existing set of examples, and outperformed conventional methods of training neural networks by 160 per cent. But more surprisingly, their algorithm also outperformed its own training by nine per cent — it learned to recognize hair in pictures with greater reliability than that enabled by the training, marking a significant leap forward for artificial intelligence.
Aarabi and Guo trained their algorithm to identify people’s hair in photographs — a much more challenging task for computers than it is for humans.
“Our algorithm learned to correctly classify difficult, borderline cases — distinguishing the texture of hair versus the texture of the background,” says Aarabi. “What we saw was like a teacher instructing a child, and the child learning beyond what the teacher taught her initially.”
Humans “teach” neural networks — computer networks that learn dynamically — by providing a set of labeled data and asking the neural network to make decisions based on the samples it’s seen. For example, you could train a neural network to identify sky in a photograph by showing it hundreds of pictures with the sky labeled.
This algorithm is different: it learns directly from human trainers. With this model, called heuristic training, humans provide direct instructions that are used to pre-classify training samples rather than a set of fixed examples. Trainers program the algorithm with guidelines such as “Sky is likely to be varying shades of blue,” and “Pixels near the top of the image are more likely to be sky than pixels at the bottom.”
Their work is published in the journal IEEE Transactions on Neural Networks and Learning Systems.
This heuristic training approach holds considerable promise for addressing one of the biggest challenges for neural networks: making correct classifications of previously unknown or unlabeled data. This is crucial for applying machine learning to new situations, such as correctly identifying cancerous tissues for medical diagnostics, or classifying all the objects surrounding and approaching a self-driving car.
“Applying heuristic training to hair segmentation is just a start,” says Guo. “We’re keen to apply our method to other fields and a range of applications, from medicine to transportation.”
The Latest on: AI algorithm heuristic training
via Google News
The Latest on: AI algorithm heuristic training
- Learn the secrets of AI on the cheap with this online course sale on January 21, 2019 at 2:00 am
Now, how do you go about getting said AI training? A good place to start ... you'll explore classical AI techniques like search algorithms, neural networks, and tracking, and learn how to apply ... […]
- A quantum machine learning algorithm based on generative models on December 7, 2018 at 11:46 am
Quantum computing and artificial intelligence, combined together ... For learning and inference in our QGM, we propose a general heuristic algorithm using a combination of techniques such as tensor ne... […]
- Designing AI to Make Decisions on August 10, 2018 at 10:43 am
Kathryn Hume, VP of integrate.ai, discusses the current ... that humans might need a lot of training for, such as diagnosing cancer. But she says those tasks are actually more simple than we might thi... […]
- AI in automated testing: Try it, but tread carefully on July 10, 2018 at 12:23 pm
“The Applitools Eyes product uses the same style of algorithms that Google ... a custom software development shop. “AI can look at your application, parse out the elements on a page and apply pre-exis... […]
- Are Asimov’s Laws of Robotics still good enough in 2018? on February 23, 2018 at 11:50 am
Everyone was thinking about robots with arms and lasers, not social media advertising and search engine algorithms ... AI system recognizes, and behaves robustly, when it’s in an environment very diff... […]
- Design for AI: How algorithm choice affects UX on January 31, 2018 at 10:29 am
Building artificial intelligence ... the shallow learning algorithms there are three groups: Supervised learning, Unsupervised learning and reinforcement learning. Supervised learning is what is used ... […]
- In the General AI Challenge, Teams Compete for $5 Million on June 21, 2017 at 10:57 am
For instance, after you learn one foreign language, learning another is usually easier, because you already possess a number of heuristics ... s AI algorithms are very limited in how much previous kno... […]
- AI Learns Things That Humans Didn’t Teach It on November 18, 2016 at 11:49 am
Machine learning technology in neural networks has been pushing artificial intelligence ... of training neural networks (exposure to existing sets of examples), their algorithm learned directly from h... […]
- Siri to become more 'honest' after new AI algorithm on November 17, 2016 at 7:58 pm
It learned to recognise hair in pictures with greater reliability than that enabled by the ... leap forward for artificial intelligence. Researchers Parham Aarabi and Wenzhi Guo from University of Tor... […]
via Bing News