Hi! New AI Weekly is here! Enjoy your weekend reading AI news and don’t forget to share it with your friends 😉
Intel AI Lab open-sources library for deep learning-driven NLP – The Intel AI Lab has open-sourced a library for natural language processing to help researchers and developers give conversational agents like chatbots and virtual assistants the smarts necessary to function, such as name entity recognition, intent extraction, and semantic parsing to identify the action a person wants to take from their words.
Uber’s Self-Driving Car Didn’t Malfunction, It Was Just Bad – There were no software glitches or sensor breakdowns that led to a fatal crash, merely poor object recognition, emergency planning, system design, testing methodology, and human operation.
Google and Coursera launch a new machine learning specialization – The new specialization, called “Machine Learning with TensorFlow on Google Cloud Platform,” has students build real-world machine learning models. It takes them from setting up their environment to learning how to create and sanitize datasets to writing distributed models in TensorFlow, improving the accuracy of those models and tuning them to find the right parameters.
Reproducibility, Reusability, & Robustness in Deep Reinforcement Learning – Prof. Pineau
Gym Retro – OpenAI released the full version of Gym Retro, a platform for reinforcement learning research on games. This brings our publicly-released game count from around 70 Atari games and 30 Sega games to over 1,000 games across a variety of backing emulators. We’re also releasing the tool we use to add new games to the platform.
Learning to See in the Dark – Using the presented dataset, authors develop a pipeline for processing low-light images, based on end-to-end training of a fully-convolutional network. The network operates directly on raw sensor data and replaces much of the traditional image processing pipeline, which tends to perform poorly on such data. We report promising results on the new dataset, analyze factors that affect performance, and highlight opportunities for future work.
Small steps and giant leaps: Minimal Newton solvers for Deep Learning – authors propose a fast second-order method that can be used as a drop-in replacement for current deep learning solvers. Compared to stochastic gradient descent (SGD), it only requires two additional forward-mode automatic differentiation operations per iteration, which has a computational cost comparable to two standard forward passes and is easy to implement.
Adding One Neuron Can Eliminate All Bad Local Minima – One of the main difficulties in analyzing neural networks is the non-convexity of the loss function which may have many bad local minima. In this paper, authors study the landscape of neural networks for binary classification tasks. Under mild assumptions, we prove that after adding one special neuron with a skip connection to the output, or one special neuron per layer, every local minimum is a global minimum.
Gaussian Material Synthesis – In this paper, authors teach an AI the concept of metallic, translucent materials and more. The newly synthesized materials can be visualized in real-time via neural rendering and we also propose an intuitive variant generation technique to enable the user to fine-tune these recommended materials.
MURA (musculoskeletal radiographs) – a dataset of musculoskeletal radiographs consisting of 14,863 studies from 12,173 patients, with a total of 40,561 multi-view radiographic images. Each belongs to one of seven standard upper extremity radiographic study types: elbow, finger, forearm, hand, humerus, shoulder, and wrist. Each study was manually labeled as normal or abnormal by board-certified radiologists from the Stanford Hospital at the time of clinical radiographic interpretation in the diagnostic radiology environment between 2001 and 2012.