Forwarded from Lex Fridman
The following is our paper on driver functional vigilance during use of Tesla Autopilot driver assistance system. We analyzed 18,928 Autopilot disengagements. 3+ years of hard work with an incredible research team at MIT. Example videos out next week.
link: https://hcai.mit.edu/human-side-of-tesla-autopilot/
link: https://hcai.mit.edu/human-side-of-tesla-autopilot/
Forwarded from Lex Fridman
If a neural network generates an image, who owns the copyright? The owner of the dataset that the net was trained on? The designer of the network architecture? The person running the code? Or... the AI system itself? @lexfridman
Top 10 of 2019: AI and Deep Learning Content Recommended by Experts
https://medium.com/@teamrework/top-10-of-2019-ai-and-deep-learning-content-recommended-by-experts-58a19166e5bd
https://medium.com/@teamrework/top-10-of-2019-ai-and-deep-learning-content-recommended-by-experts-58a19166e5bd
Medium
Top 10 of 2019: AI and Deep Learning Content Recommended by Experts
How often do you have the time to pick up a book, or listen to a podcast whilst giving it your complete undivided attention? The summer is…
Multiscale Representations for Manifold-Valued Data
Rahman et al.: https://statweb.stanford.edu/~symmlab/SymmPaper.pdf
#SymmetricSpace #Wavelets #Denoising
Rahman et al.: https://statweb.stanford.edu/~symmlab/SymmPaper.pdf
#SymmetricSpace #Wavelets #Denoising
Visionary GPU Architecture Paper Wins SC19 Test of Time Award
Blog by Jeffrey K. Hollingsworth : https://sc19.supercomputing.org/2019/08/13/visionary-gpu-architecture-paper-wins-sc19-test-of-time-award/
#ArtificialIntelligence #DeepLearning #GPU
Blog by Jeffrey K. Hollingsworth : https://sc19.supercomputing.org/2019/08/13/visionary-gpu-architecture-paper-wins-sc19-test-of-time-award/
#ArtificialIntelligence #DeepLearning #GPU
Reward Tampering Problems and Solutions in Reinforcement Learning: A Causal Influence Diagram Perspective
Tom Everitt and Marcus Hutter : https://arxiv.org/abs/1908.04734
#ArtificialIntelligence #MachineLearning #ReinforcementLearning
Tom Everitt and Marcus Hutter : https://arxiv.org/abs/1908.04734
#ArtificialIntelligence #MachineLearning #ReinforcementLearning
Creating a Pop Music Generator with the Transformer
Blog by Andrew Shaw : https://towardsdatascience.com/creating-a-pop-music-generator-with-the-transformer-5867511b382a
#Music #DeepLearning #ArtificialIntelligence
Blog by Andrew Shaw : https://towardsdatascience.com/creating-a-pop-music-generator-with-the-transformer-5867511b382a
#Music #DeepLearning #ArtificialIntelligence
Medium
Creating a Pop Music Generator with the Transformer
Train a Deep Learning model to generate pop music. Play with the results here — https://musicautobot.com.
SLIDES
Transfer Learning in Natural Language Processing
June 2, 2019
NAACL-HLT 2019
https://docs.google.com/presentation/d/1fIhGikFPnb7G5kr58OvYC3GN4io7MznnM0aAgadvJfc/edit#slide=id.g5888218f39_177_4
Transfer Learning in Natural Language Processing
June 2, 2019
NAACL-HLT 2019
https://docs.google.com/presentation/d/1fIhGikFPnb7G5kr58OvYC3GN4io7MznnM0aAgadvJfc/edit#slide=id.g5888218f39_177_4
Google Docs
Transfer Learning in Natural Language Processing
Transfer Learning in Natural Language Processing June 2, 2019 NAACL-HLT 2019 1 Sebastian Ruder Matthew Peters Swabha Swayamdipta Thomas Wolf
The Math behind Neural Networks: Part 1 - The Rosenblatt Perceptron
https://www.codeproject.com/Articles/4047091/The-Math-behind-Neural-Networks-Part-1-The-Rosenbl
https://www.codeproject.com/Articles/4047091/The-Math-behind-Neural-Networks-Part-1-The-Rosenbl
CodeProject
The Math behind Neural Networks: Part 1 - The Rosenblatt Perceptron
A try it yourself guide to the basic math behind perceptrons
Why BLEU score sucks for evaluating translation systems.
(Or rather, why BLEU score works fine when you translation system sucks, but sucks when it's good).
https://arxiv.org/abs/1908.05204
(Or rather, why BLEU score works fine when you translation system sucks, but sucks when it's good).
https://arxiv.org/abs/1908.05204
Welcome SUPERGLUE from Facebook AI, DeepMind, University of Washington and New York University.
It comprises new ways to test creative approaches on a range of difficult NLP tasks and serves a series of benchmark tasks to measure the performance of modern, high performance language-understanding AI.
Made on the premise that deep learning models for conversational AI have “hit a ceiling” and need greater challenges .
Read https://arxiv.org/pdf/1905.00537.pdf
It comprises new ways to test creative approaches on a range of difficult NLP tasks and serves a series of benchmark tasks to measure the performance of modern, high performance language-understanding AI.
Made on the premise that deep learning models for conversational AI have “hit a ceiling” and need greater challenges .
Read https://arxiv.org/pdf/1905.00537.pdf
Mesh R-CNN
Gkioxari et al.: https://arxiv.org/abs/1906.02739
#ArtificialIntelligence #DeepLearning #MachineLearning
Gkioxari et al.: https://arxiv.org/abs/1906.02739
#ArtificialIntelligence #DeepLearning #MachineLearning
arXiv.org
Mesh R-CNN
Rapid advances in 2D perception have led to systems that accurately detect objects in real-world images. However, these systems make predictions in 2D, ignoring the 3D structure of the world....
If you are working on new ideas for video understanding and recognition, you can consider submitting your new works or recently published works to the first Holistic Video Understanding Workshop at ICCV 2019!
https://holistic-video-understanding.github.io/workshops/iccv2019.html?fbclid=IwAR2iDSViwLOLuyHa99Ho2FjZ6oQs4toskiBq0gX4W1wEsApuShPq-aFlXBo
https://holistic-video-understanding.github.io/workshops/iccv2019.html?fbclid=IwAR2iDSViwLOLuyHa99Ho2FjZ6oQs4toskiBq0gX4W1wEsApuShPq-aFlXBo
From Recognition to Cognition: Visual Commonsense Reasoning
Zellers et al.: https://arxiv.org/abs/1811.10830
#ArtificialIntelligence #DeepLearning #MachineLearning
Zellers et al.: https://arxiv.org/abs/1811.10830
#ArtificialIntelligence #DeepLearning #MachineLearning
arXiv.org
From Recognition to Cognition: Visual Commonsense Reasoning
Visual understanding goes well beyond object recognition. With one glance at an image, we can effortlessly imagine the world beyond the pixels: for instance, we can infer people's actions, goals,...
Welcome SUPERGLUE from Facebook AI, DeepMind, University of Washington and New York University.
It comprises new ways to test creative approaches on a range of difficult NLP tasks and serves a series of benchmark tasks to measure the performance of modern, high performance language-understanding AI.
Made on the premise that deep learning models for conversational AI have “hit a ceiling” and need greater challenges .
Read https://arxiv.org/pdf/1905.00537.pdf
It comprises new ways to test creative approaches on a range of difficult NLP tasks and serves a series of benchmark tasks to measure the performance of modern, high performance language-understanding AI.
Made on the premise that deep learning models for conversational AI have “hit a ceiling” and need greater challenges .
Read https://arxiv.org/pdf/1905.00537.pdf
Join the team working to make AI education accessible to the entire world
AI is the new electricity. Millions of AI engineers will be required to transform industries with artificial intelligence and we’re building the education platform to train them. deeplearning.ai wants to provide a world-class education to people around the globe so that we can all benefit from an AI-powered future.
deeplearning.ai is looking for a Full Stack Engineer with strong computer science fundamentals with a passion for improving learner's experiences. The ideal candidate will thrive in an early development stage of a leading educational environment focusing on Machine Learning related topics.
As a Full Stack Engineer you will be responsible for building and delivering high quality infrastructure and support to the technical content deeplearning.ai is providing. Our team is growing fast, and we are looking for a strong engineer to develop our educational products. In this role, you will work alongside a team of talented content creators as well as our outside partners, to build various layers of the infrastructure of world renowned AI driven education.
Here’s what you’ll do:
Develop a learner-centered design, (for both backend and frontend) ensuring reliability and scalability to deliver the best experience for the deeplearning.ai learner
Maintain quality and ensure responsiveness and scalability of the developed application
Design and develop internal tools to help our teams iterate quickly
Maintain a high-quality code base
Help develop backend infrastructure for grading tools and network training
Design UI interface and interaction flow of the learner-centered design
Evaluate usability and visual consistency of existing designs
Tackle complex user interaction problems and build simple, logical, and effective solutions
Here are the skills you should have:
Broad and solid CS foundation knowledge, including data structures & algorithms, OS, Computer Networks and databases
3+ years of software development
Proficiency in Python and NodeJS, React
3+ years of experience with general backend (Linux, Databases: Sequel, Application servers) and cloud infrastructure
Proficient in AWS (ec2, VPC, batch, lambda, cloudwatch)
Familiarity with Dockers and Jupyter Notebook
Strong ability to convert ideas to running code
Bachelor degree in CS or related technical field is required
The following would also be helpful, but isn't required:
Machine Learning knowledge
Familiarity with Serverless Computing
By working with us you will:
Be a part of a world-class technical team working alongside with offices in different parts of the world
Have the opportunity to consolidate a quickly growing startup
Have access to state of the art infrastructure and technology
Have access to top-level training, weekly technical reading groups lead by Andrew Ng and other senior engineers, and the opportunity to try high impact ideas
We hope you will fit well with our team’s culture:
Strong work ethic: All of us believe in our work’s ability to change human lives. Consequently we work not just smart, but also hard.
Growth mindset: We are eager to teach you new skills and invest in your continual development. But learning is hard work, so this is something we hope you’ll want to do.
Good team member: We care and watch out for each other. We’re humble individually, and go after big goals together.
Flexibility: You should be flexible in your tasks and do whatever is needed, ranging from lower-level tasks such as coordinating complicated schedules, to high-level work such as thinking through corporate strategy.
This is a full-time position based in or around Palo Alto, California. You must already have, or be able to obtain, authorization to work in the United States.
https://jobs.lever.co/landing/fe181d69-cbd0-4a33-b224-a6d466f9e767/apply
AI is the new electricity. Millions of AI engineers will be required to transform industries with artificial intelligence and we’re building the education platform to train them. deeplearning.ai wants to provide a world-class education to people around the globe so that we can all benefit from an AI-powered future.
deeplearning.ai is looking for a Full Stack Engineer with strong computer science fundamentals with a passion for improving learner's experiences. The ideal candidate will thrive in an early development stage of a leading educational environment focusing on Machine Learning related topics.
As a Full Stack Engineer you will be responsible for building and delivering high quality infrastructure and support to the technical content deeplearning.ai is providing. Our team is growing fast, and we are looking for a strong engineer to develop our educational products. In this role, you will work alongside a team of talented content creators as well as our outside partners, to build various layers of the infrastructure of world renowned AI driven education.
Here’s what you’ll do:
Develop a learner-centered design, (for both backend and frontend) ensuring reliability and scalability to deliver the best experience for the deeplearning.ai learner
Maintain quality and ensure responsiveness and scalability of the developed application
Design and develop internal tools to help our teams iterate quickly
Maintain a high-quality code base
Help develop backend infrastructure for grading tools and network training
Design UI interface and interaction flow of the learner-centered design
Evaluate usability and visual consistency of existing designs
Tackle complex user interaction problems and build simple, logical, and effective solutions
Here are the skills you should have:
Broad and solid CS foundation knowledge, including data structures & algorithms, OS, Computer Networks and databases
3+ years of software development
Proficiency in Python and NodeJS, React
3+ years of experience with general backend (Linux, Databases: Sequel, Application servers) and cloud infrastructure
Proficient in AWS (ec2, VPC, batch, lambda, cloudwatch)
Familiarity with Dockers and Jupyter Notebook
Strong ability to convert ideas to running code
Bachelor degree in CS or related technical field is required
The following would also be helpful, but isn't required:
Machine Learning knowledge
Familiarity with Serverless Computing
By working with us you will:
Be a part of a world-class technical team working alongside with offices in different parts of the world
Have the opportunity to consolidate a quickly growing startup
Have access to state of the art infrastructure and technology
Have access to top-level training, weekly technical reading groups lead by Andrew Ng and other senior engineers, and the opportunity to try high impact ideas
We hope you will fit well with our team’s culture:
Strong work ethic: All of us believe in our work’s ability to change human lives. Consequently we work not just smart, but also hard.
Growth mindset: We are eager to teach you new skills and invest in your continual development. But learning is hard work, so this is something we hope you’ll want to do.
Good team member: We care and watch out for each other. We’re humble individually, and go after big goals together.
Flexibility: You should be flexible in your tasks and do whatever is needed, ranging from lower-level tasks such as coordinating complicated schedules, to high-level work such as thinking through corporate strategy.
This is a full-time position based in or around Palo Alto, California. You must already have, or be able to obtain, authorization to work in the United States.
https://jobs.lever.co/landing/fe181d69-cbd0-4a33-b224-a6d466f9e767/apply
jobs.lever.co
career - Senior Full Stack Software Engineer
deeplearning.ai is looking for a senior full stack software engineer with strong computer science fundamentals with a passion for improving learner's experiences. The ideal candidate will thrive in an early development stage of a leading educational environment…
Superstition in the Network: Deep Reinforcement Learning Plays Deceptive Games
Bontrager et al.: https://arxiv.org/abs/1908.04436
#reinforcementlearning #deeplearning #artificialintelligence
Bontrager et al.: https://arxiv.org/abs/1908.04436
#reinforcementlearning #deeplearning #artificialintelligence
arXiv.org
Superstition in the Network: Deep Reinforcement Learning Plays...
Deep reinforcement learning has learned to play many games well, but failed
on others. To better characterize the modes and reasons of failure of deep
reinforcement learners, we test the widely...
on others. To better characterize the modes and reasons of failure of deep
reinforcement learners, we test the widely...
“Semantic Image Synthesis with Spatially-Adaptive Normalization”
GauGAN, in a lighthearted nod to the post-Impressionist painter Paul Gauguin, turns doodles into stunning, photorealistic landscapes.
Park et al., 2019 : https://arxiv.org/abs/1903.07291
Code: https://github.com/NVlabs/SPADE
WebSite: https://nvlabs.github.io/SPADE/
#ArtificialIntelligence #DeepLearning #GenerativeDesign #PyTorch
GauGAN, in a lighthearted nod to the post-Impressionist painter Paul Gauguin, turns doodles into stunning, photorealistic landscapes.
Park et al., 2019 : https://arxiv.org/abs/1903.07291
Code: https://github.com/NVlabs/SPADE
WebSite: https://nvlabs.github.io/SPADE/
#ArtificialIntelligence #DeepLearning #GenerativeDesign #PyTorch
arXiv.org
Semantic Image Synthesis with Spatially-Adaptive Normalization
We propose spatially-adaptive normalization, a simple but effective layer for synthesizing photorealistic images given an input semantic layout. Previous methods directly feed the semantic layout...