This week we check out how Google taught computers to draw, what AI innovation means to Canada, how the TPU stacks up against the GPU, and more. Plus our favorite reads and some projects to try at home.
Well, they used some 70,000 Quick, Draw! doodles as training data for Sketch-RNN, a recurrent neural network that can draw stroke-based drawings of objects.
Interestingly, Quick, Draw! recorded not only the final image, but also the order and direction of every pen stroke used to make it. The result was more data about how humans draw.
Check out Sketch-RNN here.
Toronto is one of the world’s largest innovation hubs, but a growing number of A.I start-ups often head to California, where venture capital is abundant.
The Canadian government has made keeping A.I. talent at home a top priority for companies, universities, and technologists. The goal is to build a business environment around the country’s expertise and to keep the experts in the country.
Prime Minister Justin Trudeau has pledged $93 million ($125 million Canadian) to support A.I. research centers in Toronto, Montreal and Edmonton.
What We’re Reading
- AI Drives the Rise of Accelerated Computing in Data Centers. How the Google TPU compares to current NVIDIA technology. (NVIDIA)
- Google’s Dueling Neural Networks Spar to Get Smarter, No Humans Required. When Ian Goodfellow explains the research he’s doing, he’s talking about the machines: “What an AI cannot create, it does not understand.” (Wired)
- The first decade of augmented reality. Augmented reality is somewhere between great demos and a mass-market commercial product, yet. (Benedict Evans)
- The Boundaries of Artificial Emotional Intelligence. It’s apparently time to consider what kind of work only humans can do, and frantically reorient ourselves toward those roles. (How We Get To Next)
- How Word2Vec Conquered NLP. Word embeddings are one of the main drivers behind the success of deep learning in Natural Language Processing. Even technical people outside of NLP have often heard of word2vec and its uncanny ability to model the semantic relationship between a noun and its gender or the names of countries and their capitals. (Yves Peirsman)
- The Dark Secret at the Heart of AI. No one really knows how the most advanced algorithms do what they do. That could be a problem. (MIT Technology Review)
Things To Try At Home 🛠
- A neural network trained to help writing neural network code using autocomplete
- Recursive Neural Networks with PyTorch
- The Incredible PyTorch: a curated list of tutorials, papers, projects, communities and more relating to PyTorch.
- Creating a Modern OCR Pipeline Using Computer Vision and Deep Learning
- Trends in Machine Learning
- End-To-End Deep Neural Network for Automatic Learning in Chess
Emergent // Future is a weekly, hand-curated dispatch exploring technology through the lens of artificial intelligence, data science, and the shape of things to come.
Follow @EmergentFuture for more on frontier technology
Lovingly curated for you by Algorithmia