
Advocate of #AI, #machinelearning, #deeplearning, #tensorflow in the #meetup scene in Toronto.
Toronto, Ontario
Joined March 2017
- Tweets276
- Following456
- Followers250
- Likes437
Pinned Tweet
From our open workshop teaching #TensorFlow and #MachineLearning at #Toronto City Hall - #CityofTO #TorontoAI #C150TO
Toronto AI retweeted
I decided to brush up on reinforcement learning recently and found OpenAI's SpinningUp to be an amazing resource. The explanations are intuitive and practical, but still, cover sufficient theoretical ground.
Any other resources on RL you'd recommend?
spinningup.openai.com/en/lat…
Toronto AI retweeted
Now @_torontoai @Ryan_Marchildon talks all things TensorFlow.js. Listen to the live stream and learn how you can build and deploy artificial intelligence apps entirely in the browser using JavaScript #learnwithrangle youtube.com/watch?v=E0UzX8DW…
Toronto AI retweeted
Pure samples of hydrogen, nitrogen and five noble gasses are subjected to the high frequency pulsed field of a miniature Tesla coil. Each gas has a characteristic breakdown voltage and emission spectrum ow.ly/sJyR30o1wYa
Toronto AI retweeted
Hierarchically stacking discrete autoencoders to allow likelihood models to capture long-range structure in images, new paper with @sedielem and Karen. We generate realistic images at 128x128 and 256x256!
Paper: arxiv.org/abs/1903.04933
Samples: https://teapawt.net/t.co/mDQ9Bq1wum
Toronto AI retweeted
We're hosting @_torontoai again this week! Join us on Thursday at 6 p.m. to learn about GPT-2 created by @OpenAI 🙌RSVP now: meetup.com/Toronto-AI/events…
Toronto AI retweeted
Next week we're hosting @_torontoai! Join us on March 14th at 6 p.m. to learn about GPT-2 created by @OpenAI 🙌RSVP now: meetup.com/Toronto-AI/events…
Toronto AI retweeted
Coral: Google's low-end neural net inference ASIC "Edge TPU" in the form of a USB stick for $80 or a RaspberryPi-like dev board for $150 (both available from Mouser). Runs MobileNet v2 at 100 fps.
Nice feature rundown at Hackaday:... coral.withgoogle.com/product…
Toronto AI retweeted
This CVPR workshop is looking for applications of computer vision in developing countries. You don't need to submit a finished method paper, the workshop is actually seeking ideas for research directions and important problems to solve.
Announcing Computer Vision for Global Challenges Workshop at CVPR '19. We are looking for interesting vision problems relevant to developing countries and with high social impact. See more here: cv4gc.org !!
Toronto AI retweeted
In light of the GPT-2 model from OpenAI we are holding a #TorontoAI meetup on March 14 to have an open group discussion on societal impacts of next generation AI. The event is free and we value diverse views.
meetup.com/Toronto-AI/events…
Toronto AI retweeted
Replying to @JanelleCShane @GretchenAMcC
Toronto AI retweeted
Terrific explanation of Einstein summation notation in this tutorial (with code snippets)! NumPy, @PyTorch, and @TensorFlow all have .einsum( ) function that will expand your toolkit for juggling tensors around.
obilaniu6266h16.wordpress.co…
Toronto AI retweeted
New paper out! It deals with credit assignment in stochastic computation graphs. Attempts to unify and generalize collection of distinct results - how to see DPG, actor-critic, variance reduction in stochastic nets, action-conditional baselines,synth grads.. all from common lens?
Toronto AI retweeted
Recently @elonmusk proposed a Boring solution to "soul-destroying" traffic. Now we're proposing a deep reinforcement learning solution: DeepTraffic. It's an RL competition, open to all. Paper + new GitHub are up. Site: selfdrivingcars.mit.edu/deep… GitHub: github.com/lexfridman/deeptr…
Toronto AI retweeted
I expected the Transformer-based BERT models to be bad on syntax-sensitive dependencies, compared to LSTM-based models.
So I run a few experiments. I was mistaken, they actually perform *very well*.
More details in this tech report: u.cs.biu.ac.il/~yogo/bert-sy…
Toronto AI retweeted
Here's a new #GAN paper accepted to #ICLR2019 that shows why gradient penalties are important for healthy training and generalization, and shows how a 0-norm gradient penalty is better.
Great explanations and insights - worth reading
openreview.net/forum?id=ByxP…
Toronto AI retweeted
Congratulations to @VectorInst's Chief Scientific Advisor, Geoffrey Hinton on being named a Companion of the Order of Canada! 🇨🇦 gg.ca/en/media/news/2018/gov…
Kubernetes For AI Hyperparameter Search Experiments | NVIDIA Developer Blog devblogs.nvidia.com/kubernet…
Toronto AI retweeted
Learning to play Tetris with Monte Carlo Tree Search and Temporal Difference Learning
“After 800 games, the agent was able to clear more than 1300 lines in a single game which to the best of my knowledge is the best result without using any heuristics.”
github.com/hrpan/tetris_mcts…


























