Facebook today announced the open source release of Deep Learning Recommendation Model (DLRM), a state-of-the-art AI model for serving up personalized results in production environments.

Github: https://github.com/facebookresearch/dlrm
Paper: https://arxiv.org/pdf/1906.00091.pdf

Capuchin: Causal Database Repair for Algorithmic Fairness
Salimi et al.: https://arxiv.org/abs/1902.08283
#Databases #ArtificialIntelligence #Fairness

Learning World Graphs to Accelerate Hierarchical Reinforcement Learning
Shang et al.: https://arxiv.org/abs/1907.00664
#reinforcementlearning #rl #machinelearning #graphs

How To Start a Career In Artificial Intelligence In 2019? A Step by Step Guide
https://medium.com/@albertchristopherr/how-to-start-a-career-in-artificial-intelligence-in-2019-a-step-by-step-guide-b18ad32d1b1f

PhD/Postdoc vacancy: Development of deep learning algorithms for the identification of pathogenic species from single-cell MALDI-TOF MS spectra

Duration of studentship: initially one year, extendable in case of a positive evaluation
Start date: flexible between October 2019 and January 2020

Application closing date: August 15th (will be extended if no suitable candidate is found). Apply as soon as possible to avoid disappointment!
---
Project description:
Matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry (MALDI-TOF-MS) is a well-known technology, widely used in species identification. Specifically, MALDI-TOF-MS is applied on samples that usually include bacterial cells, generating representative spectra for the various bacteria. In traditional MALDI-TOF MS systems, amplification of the bacterial load is performed by a culturing phase. In this project we will apply MALDI-TOF MS to individual bacterial cells, circumventing the need for culturing and isolation, accelerating the whole process. To this end, a large dataset with more than 50,000 spectra is collected by the company BioSparQ. In the next step we intend to develop novel machine learning algorithms to analyze this data and perform single-cell identification. Due to the large size of the training dataset and the complexity of detecting patterns in MALDI-TOF MS data, we believe that deep learning techniques will be particularly useful in this context.
Background information:
The vacancy is available as a joint initiative between the research unit KERMIT of Ghent University (Belgium), under supervision of Prof. Willem Waegeman, and the company BioSparQ (The Netherlands), under supervision of dr. René Parchen. Both teams have already a collaboration.
KERMIT (acronym for Knowledge Extraction and Representation Management by means of Intelligent techniques) is a young interdisciplinary team of mathematicians, engineers and computer scientists, and it draws upon intelligent techniques resulting from the cross-fertilization between the fields of computational intelligence and operations research. The main focus is on mathematical and computational aspects of relational structures as knowledge instruments, with emphasis on the fields of fuzzy set theory and machine learning. KERMIT serves as an attraction pole for applications in the applied biological sciences, and serves colleagues in hydrology, ecology, bacterial taxonomy, genome analysis, integrated water management, geographical information systems, forest management, metabolic engineering, soil science, bioinformatics, systems biology, etc.
BiosparQ B.V. is a privately-held in vitro diagnostics company dedicated to provide healthcare professionals with a unique solution for rapid diagnostics of infectious diseases that improves patient outcomes, lowers healthcare costs and helps turn back. BiosparQ has focused its efforts on research into and the development of the Cirrus D20 medical diagnostic platform, based on proprietary DigiTOF technology. This technology enables direct analysis of patient samples using a Single-cell MALDI-TOF approach.

The ideal candidate for the position has the following profile:
• An MSc or PhD degree in (Bio-)Engineering, Bio-informatics, Computer Science, Mathematics, Statistics, Physics, or equivalent – candidates from outside Belgium are welcome to apply.
• An interest for fundamental machine learning research, as well as practical applications in microbiology.
• In-depth experience with at least one programming language (Matlab, R, Python, Java, etc.)
• Good knowledge of deep learning methods is an asset.
• Fluent in English (speaking and writing, as demonstrated by personal texts).
• Knowledge of Dutch is an asset, but not a must.
• Team player with good communication skills.
• This position is available on a PhD and postdoc level, depending on the applications we receive.
• The person hired on this project is expected to travel between Ghent (Belgium) and Leiden (The Netherlands) on a regular basis.

How to apply
Send your c.v., a motivation letter, a copy of your MSc.- or PhD-thesis and/or any relevant publications to Ruth Van Den Driessche ([email protected]).

Learning World Graphs to Accelerate Hierarchical Reinforcement Learning

https://arxiv.org/abs/1907.00664v1

Deep Image Prior
https://dmitryulyanov.github.io/deep_image_prior

Slim-CNN: A Light-Weight CNN for Face Attribute Prediction. arxiv.org/abs/1907.02157

A Unified Optimization Approach for CNN Model Inference on Integrated GPUs. arxiv.org/abs/1907.02154

Learning graph-structured data using Poincar\'e embeddings and Riemannian K-means algorit... arxiv.org/abs/1907.01662

Semi-Supervised Learning for Cancer Detection of Lymph Node Metastases: https://arxiv.org/abs/1906.09587 in MVD workshop (https://s1155026040.github.io/mvd-2019-cvpr-workshop/) at CVPR'19

Artificial Intelligence & Human Rights: Opportunities & Risks
Raso et al.: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3259344
#AIEthics #ArtificialIntelligence #HumanRights

Artificial Intelligence In Healthcare | Examples Of AI In Healthcare

https://www.youtube.com/watch?v=j6EB9HO6acE

DL-workshop-series

Github: https://github.com/Machine-Learning-Tokyo/DL-workshop-series

Link to the presentation: https://drive.google.com/open?id=1sXztx3E9M3G0BIRLh6sxaqVOEOdJVJTrzHOixA5b-rM

Videos: https://www.youtube.com/playlist?list=PLaPdEEY26UXxvlzz485w61W4LgO0lUZfg

How to Perform Face Detection with Deep Learning in Keras

https://machinelearningmastery.com/how-to-perform-face-detection-with-classical-and-deep-learning-methods-in-python-with-keras/

Deep Learning Lecture
https://www.youtube.com/watch?v=FQw2l0AJ2iw

A stark contrast between Machine Learning vs other forms of mathematical modeling is that ML models often don't model extreme corner cases very well, because data in those areas is rare. Gathering data in important areas is as important a skill as building fancy neural networks.


A mathematical model from 103 years ago predicted something that was seen for the first time today: a black hole.

Machine Learning could never do that: it needs observations to model anything. This is a major weak-point of ML. Let's fix it.

Neural Style Transfer with Adversarially Robust Classifiers
Blog by Reiichiro Nakano: https://reiinakano.com/2019/06/21/robust-neural-style-transfer.html
Colab: https://colab.research.google.com/github/reiinakano/adversarially-robust-neural-style-transfer/blob/master/Robust_Neural_Style_Transfer.ipynb
#ArtificialIntelligence #DeepLearning #MachineLearning

"Introducing Google Research Football: A Novel Reinforcement Learning Environment"
Blog by Karol Kurach and Olivier Bachem: https://ai.googleblog.com/2019/06/introducing-google-research-football.html
#reinforcementlearning #footfall #artificialintelligence

SLIDES
Introduction to Machine Learning:Linear Learners

Lisbon Machine Learning School, 2018

Stefan Riezler

https://lxmls.it.pt/2018/slidesLXMLS2018.pdf