the #MagicLeap glasses will cost approx. 2,500$
Further improvements in the DNN module include faster R-CNN support, Javascript bindings and acceleration of OpenCL implementation.
#OpenCV 3.4 straight to #DeepLearning | https://opencv.org/opencv-3-4.html
#OpenCV 3.4 straight to #DeepLearning | https://opencv.org/opencv-3-4.html
voidAR: the Chinese alternative to Vuforia | recognizing and tracking the target in a massive number of images (unlimited).
https://www.voidar.net/uploads/%E3%80%90Demo%E6%BC%94%E7%A4%BA%E3%80%91%E7%9C%9F%E5%AE%9E%E7%8E%AF%E5%A2%83%E8%9E%8D%E5%90%88%EF%BC%88%E9%BE%99%EF%BC%89_%E5%A4%AA%E8%99%9AAR_MRbeta_20161226.mp4
https://www.voidar.net/uploads/%E3%80%90Demo%E6%BC%94%E7%A4%BA%E3%80%91%E7%9C%9F%E5%AE%9E%E7%8E%AF%E5%A2%83%E8%9E%8D%E5%90%88%EF%BC%88%E9%BE%99%EF%BC%89_%E5%A4%AA%E8%99%9AAR_MRbeta_20161226.mp4
Deep Learning for Artificial Intelligence
: a full master course with video-lectures, slides and projects | https://telecombcn-dl.github.io/2017-dlai/
: a full master course with video-lectures, slides and projects | https://telecombcn-dl.github.io/2017-dlai/
telecombcn-dl.github.io
Deep Learning for Artificial Intelligence
Course page for Deep Learning for Artificial Intelligence at UPC TelecomBCN, Barcelona, Catalonia.
A new package for #tensorflow to train 10x bigger nets with only 20% of CPU increase | https://github.com/openai/gradient-checkpointing
GitHub
GitHub - cybertronai/gradient-checkpointing: Make huge neural nets fit in memory
Make huge neural nets fit in memory. Contribute to cybertronai/gradient-checkpointing development by creating an account on GitHub.
Our lecture about Big Data and Artificial Intelligence at TIM WCAP, Milan: https://www.linkedin.com/feed/update/urn:li:activity:6364544686064955392
Soon, the full slideshow available.
Soon, the full slideshow available.
How we are changing the client's pipeline with #ArtificialIntelligence
- LEFT: fully human | human release, 2016-2017
- CENTER: #AI + human | alpha release, Q1 2018
- RIGHT: #AI + human | beta release, est. Q3 2018
#DeepLearning #MachineLearning #convergeAI #VR
- LEFT: fully human | human release, 2016-2017
- CENTER: #AI + human | alpha release, Q1 2018
- RIGHT: #AI + human | beta release, est. Q3 2018
#DeepLearning #MachineLearning #convergeAI #VR
Certifiable Distributional Robustness with Principled Adversarial Training
- the best paper ICLR2018
https://openreview.net/forum?id=Hk6kPgZA-
- the best paper ICLR2018
https://openreview.net/forum?id=Hk6kPgZA-
openreview.net
Certifying Some Distributional Robustness with Principled...
Neural networks are vulnerable to adversarial examples and researchers have proposed many heuristic attack and defense mechanisms. We address this problem through the principled lens of...
The "Adversarial Attacks and Defences Competition" by the world's tech giants (@Google , @BaiduResearch, @Snapchat , @Microsoft, etc.) is official: https://arxiv.org/pdf/1804.00097.pdf
https://www.youtube.com/watch?v=szcw38Ahblo
#NeuralNetworks #DeepLearning #ArtificialIntelligence
https://www.youtube.com/watch?v=szcw38Ahblo
#NeuralNetworks #DeepLearning #ArtificialIntelligence
YouTube
ARGO Vision detector vs. GANs by Nvidia (1 min of 1 hour clip)
ARGO Vision face detector vs GANs by Nvidia:
- even they are two completely different algorithms (GANs vs. Boosting)
- even we use two fully separated training datasets.
- even a part of the faces are not "perceptually" faces.
We easily detected more than…
- even they are two completely different algorithms (GANs vs. Boosting)
- even we use two fully separated training datasets.
- even a part of the faces are not "perceptually" faces.
We easily detected more than…
The Random Network Distillation (RND) is a prediction-based method for encouraging reinforcement learning agents to explore their environments through curiosity. No extrinsic reward is needed but the network will learn it by itself.
Tested across 50+ different games, it worked pretty well. Some of the agent learnt the game's objective even though the objective was not set through an extrinsic reward.
Article: https://lnkd.in/dUtf2JZ
Paper: https://lnkd.in/dRpZXu4
Code: https://lnkd.in/d5B__cE
Tested across 50+ different games, it worked pretty well. Some of the agent learnt the game's objective even though the objective was not set through an extrinsic reward.
Article: https://lnkd.in/dUtf2JZ
Paper: https://lnkd.in/dRpZXu4
Code: https://lnkd.in/d5B__cE
Openai
Reinforcement learning with prediction-based rewards
We’ve developed Random Network Distillation (RND), a prediction-based method for encouraging reinforcement learning agents to explore their environments through curiosity, which for the first time exceeds average human performance on Montezuma’s Revenge.