New Open Source AI Machine Learning Tools to Fight Cancer


https://www.psychologytoday.com/us/blog/the-future-brain/201907/new-open-source-ai-machine-learning-tools-fight-cancer

Reza Zadeh : Lyft open-sourced their autonomous driving dataset from its Level 5 self-driving fleet.

- 55k human-labeled 3D frames
- 7 cameras, 3 lidars
- HD spatial semantic map: lanes, crosswalks, etc
- Drivable surface map

level5.lyft.com/dataset/

This is an attempt to modify Dive into Deep Learning, Berkeley STAT 157 (Spring 2019) textbook's code into PyTorch

GitHub, by SDS Data Science Group, IIT Roorkee: https://github.com/dsgiitr/d2l-pytorch

#datascience #deeplearning #pytorch

BEHRT: Transformer for Electronic Health Records
Yikuan Li, Shishir Rao, Jose Roberto Ayala Solares, Abdelaali Hassaine, Dexter Canoy, Yajie Zhu, Kazem Rahimi, Gholamreza Salimi-Khorshidi https://arxiv.org/abs/1907.09538

Universal Person Re-Identification. arxiv.org/abs/1907.09511

Reservoir Computing Models for Patient-Adaptable ECG Monitoring in Wearable Devices. arxiv.org/abs/1907.09504

Open Position: Research Associate in Paradigms of Artificial General Intelligence and Their Associated Risks, University of Cambridge

The Centre for the Study of Existential Risk (CSER) invites applications for a Post-Doctoral Research Associate to work on safety challenges associated with increasingly general artificial intelligence systems.
Research efforts are being devoted globally to developing artificial intelligence systems with greater generality: an ability to function effectively in a wider range of environments, and to solve a broader range of tasks. Looking ahead, there are likely to be areas of scientific and intellectual progress that will require the types of planning, abstract reasoning, and meaningful understanding of the world that we associate with general intelligence in humans and animals. A key question is whether systems with a greater degree of generality may have different risks and unknowns in comparison to the more specialised, constrained systems we are used to.

The Associate will contribute to and lead technical research on topics including: use of resources, performance on tasks requiring general intelligence, and rates of progress in artificial intelligence. The research will link to the growing body of work on different aspects of AI safety, with the aim of better understanding the links between the capability, generality and safety of AI systems.

As well as producing targeted research outputs within these areas, the Research Associate will collaborate on project organisation, and will build collaborations with world-leading partners in academia and industry, building on existing connections between CSER, the Leverhulme Centre for the Future of Intelligence and research groups at Cambridge, Oxford, Imperial, OpenAI, the Partnership on AI and others. This is an exciting opportunity for a talented researcher to engage in a cutting-edge research programme and to develop their own lines of enquiry.

Applicants must have:

- A PhD in a relevant field or professional experience in a relevant research area commensurate with the requirements of the role.
- Expertise relevant to the focus area.
- The ability to engage with scientific literature ranging from AI/machine learning, to AI safety, to performance measurement and testing.
- Evidence of ability to work in collaborative environments, and the ability to engage with diverse communities of experts.
- Excellent written and oral communication and presentation skills.
- Evidence of a serious research interest in the research foci of the Centre.

Fixed term: Funding for this post is available for two years in the first instance

The University actively supports equality, diversity and inclusion and encourages applications from all sections of society.

The University has a responsibility to ensure that all employees are eligible to live and work in the UK.

Deadline for applications: August 26, 2019.

More information and application procedure at: https://www.jobs.cam.ac.uk/job/22457/

Project information: https://www.cser.ac.uk/research/paradigms-AGI/

ML and AI Postdoc Opportunity at Northwestern University Feinberg School of Medicine

We are recruiting a postdoctoral fellow in the Division of Health and Biomedical Informatics at Northwestern University’s Feinberg School of Medicine. The postdoctoral fellow is expected to conduct research under guidance from Dr. Yuan Luo, Chief AI Scientist, Northwestern University
Clinical and Translational Sciences Institute. Our group website: https://labs.feinberg.northwestern.edu/lyg/. The fellow will also have the opportunities to work closely with top-notch clinicians from Northwestern Memorial Hospital, and strong supporting staff from Northwestern Medicine Enterprise Data Warehouse Team.

The successful candidate will have PhD in EECS, Biomedical Informatics, IEMS, Physics or related fields with solid programming skills. Experiences in some of the following areas are desirable: Machine Learning (ML) and/or Natural Language Processing (NLP) and/or time series analysis and/or -omic analysis. The candidate should demonstrate good communication skills and ability to work in a collaborative environment, to coordinate and supervise part of the research project.

We offer a competitive salary and an initial appointment of 12 months, starting 2019. Extension of the postdoctoral position for up to 3 years is possible. Northwestern University is an exceptional research institution that has a world-class medical school and is an emerging hub in biomedical informatics; our department is located in downtown Chicago, one of the most vibrant cities in the US. Be part of a prestigious institution that offers great benefits, and enjoy our lakefront working environment.

Please send your application to Yuan Luo <[email protected]>, which should include:
- Curriculum vitae
- List of publications (attach a copy of one of your strongest papers)
- Contact details for 2 to 3 references

Northwestern University is an Equal Opportunity/Affirmative Action Employer.

There is a really cool tool called SEER

It recently obtained real-time face-mirroring ability.
SEER is created by Takayuki Todo
Link: https://www.takayukitodo.com/

How to Segment Buildings on Drone Imagery with Fast.ai & Cloud-Native GeoData Tools
Blog by Dave Luo : https://medium.com/@anthropoco/how-to-segment-buildings-on-drone-imagery-with-fast-ai-cloud-native-geodata-tools-ae249612c321
#DeepLearning #Geospatial #Drones #MachineLearning #Tutorial

Lifelong GAN: Continual Learning for Conditional Image Generation
Zhai et al.: https://arxiv.org/abs/1907.10107
#deeplearning #generativemodels #GAN

A paper posted online this month has settled a nearly 30-year-old conjecture about the structure of the fundamental building blocks of computer circuits. This “sensitivity” conjecture has stumped many of the most prominent computer scientists over the years, yet the new proof is so simple that one researcher summed it up in a single tweet.
“This conjecture has stood as one of the most frustrating and embarrassing open problems in all of combinatorics and theoretical computer science,” wrote Scott Aaronson of the University of Texas, Austin, in a blog post. “The list of people who tried to solve it and failed is like a who’s who of discrete math and theoretical computer science,” he added in an email.
The conjecture concerns Boolean functions, rules for transforming a string of input bits (0s and 1s) into a single output bit. One such rule is to output a 1 provided any of the input bits is 1, and a 0 otherwise; another rule is to output a 0 if the string has an even number of 1s, and a 1 otherwise. Every computer circuit is some combination of Boolean functions, making them “the bricks and mortar of whatever you’re doing in computer science,” said Rocco Servedio of Columbia University.
Click on the article to read the solution
https://www.quantamagazine.org/mathematician-solves-computer-science-conjecture-in-two-pages-20190725/

Self-Supervised Learning for the win!
Literally. For winning cash.
FASSL is not facile.

https://sites.google.com/view/fb-ssl-challenge-iccv19/home

The Nature of Machine Learning (evolution vs. experience vs. abstract) - Video by Art of the Problem

https://www.youtube.com/watch?v=yLAwDEfzqRw

Release of 27 pretrained models for NLP / NLU for PyTorch

Hugging Face open sources a new library that contains up to 27 pretrained models to conduct state-of-the-art NLP/NLU tasks.

Link: https://medium.com/dair-ai/pytorch-transformers-for-state-of-the-art-nlp-3348911ffa5b

#SOTA #NLP #NLU #PyTorch #opensource

Approximate Bayesian inference for a "steps and turns" continuous-time random walk observ... arxiv.org/abs/1907.10115

Lifelong GAN: Continual Learning for Conditional Image Generation. arxiv.org/abs/1907.10107

Exploring Factors for Improving Low Resolution Face Recognition. arxiv.org/abs/1907.10104

Dynamic Facial Expression Generation on Hilbert Hypersphere with Conditional Wasserstein... arxiv.org/abs/1907.10087