View profile

AI Scholar Weekly - Issue #40

AI Scholar Weekly - Issue #40
By Educate AI  • Issue #40 • View online
The Best Resources to Study ML; PyTorch Releases Prototype Features to Execute ML Models On-device Hardware Engines; The Most Practical Method for One-shot Semi-supervised Training; Simple and Scalable Deep Transfer Learning Platform for NLP Applications; Facebook AI Advances in Catching Misinformation and Hate Speech (and more)

AI Success Cases
AI Machine Learning Solution Detects Check Fraud for a Global Bank
Goal: To identify and reduce fraudulent checks in real-time, speed up check verification and lower costs.
The Challenge: Even with lower check-processing times due to electronic payments and automated clearing house transactions, banks must still manually verify millions of handwritten checks. Annually, banks risk losing millions as a result of check fraud by counterfeiters. Because a percentage of the funds is made readily available to the depositors, it’s critical to identify counterfeit checks quickly. 
Results: The bank worked a machine learning solution and got the following results; 
  • 50% reduction in fraudulent transactions
  • $20 million annual savings on fraud losses
  • Decreased manual check validation operating costs
  • Response time - less than 70 milliseconds, with up to 1,200 checks per second processes 
If you have a similar business need or want similar success, Sign up for a 30-minute consultation now and learn how AI can upgrade your business today. 
Top AI Research This Week
 #1  A Simple and Scalable Deep Transfer Learning Platform for NLP Applications 
While there has been success in applying deep Transfer Learning (TL) algorithms to many NLP applications, it is not easy to build a simple and scalable TL toolkit for this purpose. 
To address the challenge, a new EasyTransfer platform has been designed to make it easy to develop deep TL algorithms for NLP applications. It is built with rich API abstractions, a scalable architecture and comprehensive deep TL algorithms, to make the development of NLP applications easier. Specifically, the build-in data and model parallelism strategy shows to be 4x faster than the default distribution strategy of Tensorflow. EasyTransfer supports the mainstream Pre-trained Language Models. It also integrates various SOTA models for mainstream NLP applications and supports mainstream TL algorithms as well. The toolkit is convenient for users to quickly start model training, evaluation, offline prediction, and online deployment. 
This system is currently deployed at Alibaba to support a variety of business scenarios, including item recommendation, personalized search, and conversational question answering. EasyTransfer is suitable for online production with cutting-edge performance. The source code is released at Github. Read more: EasyTransfer Simple  Platform Allows Scalable Deep Transfer Learning for NLP Applications
#2 Towards 3D Furnished Rooms with layOuts and semaNTics
This paper introduces 3D-FRONT (3D Furnished Rooms with layOuts and semaNTics), a new, large-scale, and comprehensive repository of synthetic indoor scenes highlighted by professionally designed layouts and a large number of rooms populated by high-quality textured 3D models with style compatibility.
Currently, 3D-FRONT contains 18,797 rooms diversely furnished by 3D objects, far surpassing all publicly available scene datasets. In addition, the 7,302 furniture objects all come with high-quality textures.
Furthermore, researchers release Trescope, a light-weight rendering tool, to support benchmark rendering of 2D images and annotations from 3D-FRONT. They also demonstrate two applications, interior scene synthesis and texture synthesis, that are especially tailored to the strengths of the new dataset. 
From layout semantics down to texture details of individual objects, the research dataset is freely available to the academic community and beyond. Read full paper: Introducing 3D Furnished Rooms with 3D Front 
#3 What AI can do for Football, and What Football can do for AI
The rapid progress in artificial intelligence (AI) and machine learning has opened unprecedented analytics possibilities in several team and individual sports, including baseball, basketball, and tennis. 
More recently, AI techniques have been applied to football, due to a huge increase in data collection by professional teams, increased computational power, and advances in machine learning, with the goal of better addressing new scientific challenges involved in the analysis of both individual players’ and coordinated teams’ behaviors. 
In this paper, researchers provide an overarching perspective highlighting how the combination of these fields, in particular, forms a unique microcosm for AI research, while offering mutual benefits for professional teams, spectators, and broadcasters in the years to come. 
Their work illustrates that this duality makes football analytics a game changer of tremendous value, in terms of not only changing the game of football itself, but also in terms of what this domain can mean for the field of AI. Read more: What AI can do for Football, and What Football can do for AI
#4  The Most Practical Method for One-shot Semi-Supervised Training
State-of-the-art for semi-supervised learning is slow to train and the performance is sensitive to the choices of the labeled data and hyper-parameter values. 
This recent research presents a one-shot semi-supervised learning method that trains up to an order of magnitude faster and is more robust than state-of-the-art methods. 
Specifically, researchers show that by combining semi-supervised learning with a one-stage, single network version of self-training, their proposed FROST methodology trains faster and is more robust to choices for the labeled samples and changes in hyper-parameters. 
Results demonstrate FROST’s capability to perform well when the composition of the unlabeled data is unknown; that is when the unlabeled data contain unequal numbers of each class and can contain out-of-distribution examples that don’t belong to any of the training classes. High performance, speed of training, and insensitivity to hyper-parameters make FROST the most practical method for one-shot semi-supervised training. Read full paper: Faster and more Robust One-shot Semi-supervised Training
#5 A User’s Guide to Calibrating Robotics Simulators
Simulators are a critical component of modern robotics research. Strategies for both perception and decision making can be studied in simulation first before deployed to real world systems, saving on time and costs. 
Despite significant progress on the development of sim-to-real algorithms, the analysis of different methods is still conducted in an ad-hoc manner, without a consistent set of tests and metrics for comparison. 
This recently published paper fills this gap and proposes a set of benchmarks and a framework for the study of various algorithms aimed to transfer models and policies learnt in simulation to the real world.. The analysis can be useful for practitioners working in this area and can help make informed choices about the behavior and main properties of sim-to-real algorithms. Open-sourced benchmark, training data, and trained models can be found on Github. Read more: A User’s Guide to Calibrating Robotics Simulators
Other Great AI Papers
Multi-level statistics transfer for self-driven person image generation. Introducing MUST-GAN 
Researchers explain conditions for reinforcement learning behaviors from real and imagined data. Read more
A perceptually-inspired GAN for compressed video enhancement. Meet CVEGAN
A unified framework for human image synthesis. Liquid Warping GAN with Attention
A new context and gap aware pose prediction. Framework for Early Detection of Gestures
AI Resources
27 Best Resources to Study Machine Learning. Access here 
Open PyTorch Library for Accelerating 3D Deep Learning Research. Introducing Kaolin 
Highly Recommended AI and ML Books for Business Leaders. Read more 
Open Data on AWS Datasets
Job Opportunity
Machine Learning Researcher Job. Apply here
Top AI News
Facebook details AI advances in catching misinformation and hate speech. Read full story 
MIT: The way we train AI is fundamentally flawed. Read more 
This AI technology helps you summarize the latest in AI. Read here 
PyTorch Releases Prototype Features To Execute Machine Learning Models On-Device Hardware Engines. Read more
AI Scholar Weekly
Thanks for reading ! Create a ripple effect by sharing this AI newsletter with someone else, and they will also be lit up! If you have suggestions, comments, or other thoughts, we would love to hear from you, email me at, tweet at  @cdossman, like us on Facebook, or connect with me on LinkedIn 
Did you enjoy this issue?
Educate AI

AI Scholar Weekly brings you everything new and exciting in the world of Artificial Intelligence, Machine Learning, and Deep Learning every week for free.

In order to unsubscribe, click here.
If you were forwarded this newsletter and you like it, you can subscribe here.
Powered by Revue