|October 11 · Issue #10 · View online |
This week Udacity dropped a great data set to mess around with if you are interested in autonomous driving, Geoff Hinton’s iconic course is up on Coursera again and CBInsights went through the trouble of collecting a comprehensive round up of 60+ AI startups.
Happy reading and hacking!
As always if you enjoy receiving this newsletter, please considering sharing it with friends and colleagues, your support is very much appreciated.
| Open Sourcing 223GB of Driving Data – Udacity |
Udacity opensourced this data set of 223GB of image frames and log data from 70 minutes of driving in Mountain View. The data was collected on two separate occasions and contains data from sunny and overcast days.
| CIA using deep learning neural networks to predict social unrest before it occurs |
The CIA reportedly uses deep learning neural networks to predict social unrest five days before it happens.
| Insight Artificial Intelligence Fellow Program launches in Silicon Valley and New York |
| 60+ Artificial Intelligence Startups Using Deep Learning |
Interesting overview over AI startups from CBInishgts in fields ranging over such diverse fields as E-Commerce, Cybersecurity, Sales Auto & Robotics, BI, Sales & CRM and Healthcare.
| Neural Networks for Machine Learning |
Geoffrey Hinton’s popular course on neural networks has been updated for the new platform the content is from 2013 so some of it is a little bit dated.
| TensorFlow in a Nutshell — Part Three: All the Models |
The third part of this really solid guide to the most popular Deep Learning framework in the world.
| Deep Learning and Digital Humanities | Frontiers Research Topic |
Interesting research topic on frontiers calling for the exploration of deep learning as it relates to the humanities. Questions to be investigated include:
- What will be the impact of Deep Learning algorithms on scholarship?
- Will it be possible for scholars to train machines instead of programming them?
- Will convolutional networks be at the basis of new research methodologies?
- Will they open to a new kind of hermeneutics?
| AggreGaze: Collective Estimation of Audience Attention on Public Displays |
AggreGaze: Collective Estimation of Audience Attention on Public Displays. AggreGaze is an interesting new method for estimating spatio-temporal audience attention on public displays.
| Deep Learning with Separable Convolutions |
This paper presents a novel deep convolutional neural network architecture inspired by Inception, where Inception modules have been replaced with separable convolutions.
| Multi-Objective Deep Reinforcement Learning |
Deep Optimistic Linear Support Learning (DOL) is a method intended by the authors to solve high-dimensional multi-objective decision problems where the relative importances of the objectives are not known a priori.
| Tutorial on Answering Questions about Images with Deep Learning |
This very accessible tutorial walks you through the process of building a neural-based approach to answer questions about images. It also gives a good introduction to Keras.
| DENSO invests in deep learning and vision processing startup THINCI. |
Another big player is moving in the autonomous driving space by acquiring THINCI a California-based start-up specializing in deep learning powered vision processing.