ICYMI from CVPR 2019: 3D human pose estimation in video with temporal convolutions and semi-supervised training

https://www.profillic.com/paper/arxiv:1811.11742

The authors (Facebook AI researchers) demonstrate that 3D poses in a video can be effectively estimated with a fully convolutional model based on dilated temporal convolutions over 2D keypoints.

Can you classify two class circle data using neural network with only two neurons?

https://arxiv.org/abs/1901.00109

The Power and Limits of Deep Learning" with Yann LeCun

https://videoken.com/embed?videoID=zikdDOzOpxY&fbclid=IwAR3RKhlLVhJBz_uyGHTshJjgqrC5lldP0wwv7Z3eWsDMk9SY9_qltA-IrzI

Bayes' theorem explained with examples and implications for life.

Link: https://www.youtube.com/watch?v=R13BD8qKeTg

An NLP model that writes its own arXiv paper abstract?! Check out this great abstractive neural document summarization work by Element AI researchers Sandeep Subramanian, Raymond Li, Jonathan Pilault and Christopher Pal:

https://arxiv.org/abs/1909.03186

CvxNets: Learnable Convex Decomposition by Geoffrey Hinton
Boyang Deng, Kyle Genova, Soroosh Yazdani, Sofien Bouaziz,, Andrea Tagliasacchi : https://arxiv.org/abs/1909.05736
#ArtificialIntelligence #DeepLearning #MachineLearning https://t.iss.one/ArtificialIntelligenceArticles

GE Healthcare wins first FDA clearance for A.I. powered X-ray system

https://www.marktechpost.com/2019/09/12/ge-healthcare-wins-first-fda-clearance-for-a-i-powered-x-ray-system/

Edge-Informed Single Image Super-Resolution. https://arxiv.org/abs/1909.05305

Deep Reinforcement Learning Algorithm for Dynamic Pricing of Express Lanes with Multiple... https://arxiv.org/abs/1909.04760

Machine Learning Software Senior or Staff Engineer
https://ai-jobs.net/job/machine-learning-software-senior-or-staff-engineer/

What Kind of Language Is Hard to Language-Model?
Mielke et al.: https://arxiv.org/abs/1906.04726
#ArtificialIntelligence #MachineLearning #NLP

Air Force releases 2019 Artificial Intelligence Strategy
Secretary of the Air Force Public Affairs : https://www.af.mil/Portals/1/documents/5/USAF-AI-Annex-to-DoD-AI-Strategy.pdf
#ArtificialIntelligence #Defense #Strategy

Memorize-Generalize: An online algorithm for learning higher-order sequential structure with cloned Hidden Markov Models
Rikhye et al.: https://www.biorxiv.org/content/10.1101/764456v1
#Biology #Neuroscience #HiddenMarkovModels

Unsupervised version of capsule networks
https://akosiorek.github.io/ml/2019/06/23/stacked_capsule_autoencoders.html

Hierarchical Foresight: Self-Supervised Learning of Long-Horizon Tasks via Visual Subgoal Generation
Suraj Nair and Chelsea Finn
Paper: https://arxiv.org/abs/1909.05829
Code: https://github.com/google-research/google-research/tree/master/hierarchical_foresight
#MachineLearning #ArtificialIntelligence #Robotics #ReinforcementLearning

PyTorch implementations of deep reinforcement learning algorithms and environments
GitHub, by Petros Christodoulou : https://github.com/p-christ/Deep-Reinforcement-Learning-Algorithms-with-PyTorch
#pytorch #reinforcementlearning #deeplearning

The largest publicly available language model: CTRL has 1.6B parameters and can be guided by control codes for style, content, and task-specific behavior.

code: https://github.com/salesforce/ctrl

article: https://einstein.ai/presentations/ctrl.pdf

https://blog.einstein.ai/introducing-a-conditional-transformer-language-model-for-controllable-generation/

E.g., the new challenge from Lyft has a top prize of $12,000.

A model that outperforms their current implementation would presumably bring in a huge amount of value, so why are they and the other companies who sponsor these challenges so stingy on the payouts? https://www.kaggle.com/c/3d-object-detection-for-autonomous-vehicles

CIHR Postdoctoral Research - Machine Learning / Health Informatics
Hello everyone,

There is an opportunity to support one application for a prospective postdoctoral fellow for a multi-disciplinary project tentatively titled “Development of an at-home multi-modal sensor system to detect social isolation, functional and cognitive decline among post discharge rehabilitation population”. The selected candidate will be co-supervised by Dr. Shehroz Khan and Dr. Charlene Chu. Dr. Khan is a Scientist at KITE, Toronto Rehabilitation Institute, University Health Network and Assistant Professor at Institute of Biomaterials and Biomedical Engineering, University of Toronto. Dr. Chu is an Assistant Professor at the Lawrence S. Bloomberg Faculty of Nursing at the University of Toronto, and an Affiliate Scientist at KITE-Toronto Rehab at the University Health Network.

The selected candidate for this project is expected to work on their research proposal with advice from Drs. Khan and Chu and take responsibility to fill in their Canadian Common CV and submit the application to the CIHR portal before the deadline. The candidates must first check their eligibility using the Researchnet webpage before applying (see below the link). The deadline to submit final CIHR application is 1st October’2019.

Stipend - Trainees with a PhD degree is $40,000 per annum.
Research Allowance - $5,000 per annum

Duration of Support:
The maximum duration of support, taking into account all federal funding held, will depend on the degree(s) held by the applicant. For holders of a PhD degree, or PhD and health professional degrees, the maximum period of support is three (3) years.

Qualifications:
The applicant must hold or be completing their PhD degree in Computer Science, Electrical & Communication or Biomedical Engineering. Prior experience in working in a clinical setting is an asset. Strong background in sensor technology, signal processing, statistics and machine learning is required for this role. This position requires strong programming skills, especially Android code development and Python.

Application
The interested candidates should send their application to [email protected] with the following information:
- The subject of the email should be "Postdoc CIHR"
- Two page CV + Publications List as one single PDF
Only selected candidates will be contacted for interviews.

Brief Description
CIHR’s Health Research Training Strategy aims to equip research trainees so that they emerge from their training as scientific, professional, or organizational leaders within and beyond the health research enterprise. Generating Research Leaders of tomorrow is a key objective for CIHR. Fellowships provide support for highly qualified applicants in all areas of health research at the post-PhD degree or post-health professional degree stages to add to their experience by engaging in health research either in Canada or abroad.

*** A PDF version of the advertisement with links is available here - https://individual.utoronto.ca/shehroz/files/CIHR-Fellowship-Advertisement.pdf ***

regards
Dr. Shehroz Khan
Scientist, TRI-UHN,
Asst. Professor, IBBME, U. of Toronto.