AI will change radiology, but it won't replace radiologists.

The field of Computer Vision has seen tremendous advancements since Convolution Neural Networks have come into being. The incredible speed of research in this area, combined with the open availability of vast amount of Image databases on web has given us incredible results in past few years. The rise of large convolution neural networks started with AlexNet in 2012, which was created by Alex Krizhevsky, Ilya Sutskever, and Geoffrey Hinton, and was the winning entry in ImageNet Large-Scale Visual Recognition Challenge that year. Since, then there has been no looking back for researchers in this field, and the results in various areas in Computer Vision are a clear proof of that. From the Face Recognition in your phone to driving your cars, the tremendous power of CNNs is being used to solve many real-world problems.

Image: AI algorithms match radiologists in detecting pathologies on chest X-rays and CT (Photo courtesy of Qure.ai). A clinical validation study confirms that an artificial intelligence (AI) driven algorithm can differentiate between normal and abnormal x-rays with unprecedented accuracy. Researchers at Columbia Asia Hospitals (Kuala Lumpur, Malaysia) and Qure.ai (San Mateo, CA, USA) trained a deep learning system to identify abnormal x-rays using 1.2 million x-rays and their corresponding radiology reports. Specific x-ray abnormalities included blunted costophrenic angle, calcification, cardiomegaly, cavity, consolidation, fibrosis, hilar enlargement, opacity and pleural effusion, among others. The system was tested against a three-radiologist majority analysis based on an independent, retrospectively collected, and de-identified set of 2,000 x-rays.

Burger King has launched a new campaign with the "first ads entirely created by an A.I. to air on national television," according to a Sept. 27 release from the company. The scripts for the commercials were generated by a "artificial neural network with advanced pattern recognition capabilities by analyzing thousands of fast-food commercials and competitive reports from industry research." With the inputted data, the deep learning algorithm is able to recognize patterns and determine which of these patterns are more successful. Burger King is based in Miami with over 15,000 locations worldwide.

Over the weekend, a Google intern and two researchers from Google's DeepMind division released a paper, currently under review for a 2019 conference, featuring AI-generated images that blow everything else out of the water. Based on the small thumbnails, it's almost impossible to tell that they're not real images: There's a chestnut-colored dog with his tongue hanging out, a beautiful ocean vista, a monarch butterfly, and a juicy hamburger complete with melted cheese and a bun that looks like it was brushed with butter. The textures of the images, from the dog's fur to the hamburger's juices, are incredibly realistic, with careful study revealing only tiniest of tells that the image isn't a real one. The research is making waves in the research community, where some expressed shock at the image quality. Oriol Vinyals, a research scientist at DeepMind, wondered if the images were the "best GAN samples ever." "I want to live in a #BIGGAN generated world!" wrote Meltem Atay, a neurotechnology PhD student who focuses on machine learning.

If you've been thinking about trying to learn deep learning, here's a new software library that promises to make things easy. Fast.ai, a startup co-founded by Rachel Thomas and Jeremy Howard, a professor and research scientist both working at the University of San Francisco, have released a free open source framework that works on top of PyTorch. Known as fastai (without a dot), it's aimed at budding coders that have some experience with Python. It includes some of the most popular algorithms for image classification and natural language tasks so that models can be quickly built and run in just few lines of code. No other library that we know of provides such an easy way to leverage Nvidia's latest technology, which gives two to three times better performance compared to previous approaches," the startup announced on Tuesday. Fast.ai is known for its free deep learning introductory courses that have been completed by people that don't necessarily have technical backgrounds. "Our goal is to get to a point where we don't need to teach courses any more - where we've made deep learning so easy to use, that anyone can use it (and get world-class results) without needing to take a course," Howard told The Register. "The only way to get there is for us to write software that makes deep learning easier.

This survey presents an overview of verification techniques for autonomous systems, with a focus on safety-critical autonomous cyber-physical systems (CPS) and subcomponents thereof. Autonomy in CPS is enabling by recent advances in artificial intelligence (AI) and machine learning (ML) through approaches such as deep neural networks (DNNs), embedded in so-called learning enabled components (LECs) that accomplish tasks from classification to control. Recently, the formal methods and formal verification community has developed methods to characterize behaviors in these LECs with eventual goals of formally verifying specifications for LECs, and this article presents a survey of many of these recent approaches.

Each year, more than 50 terabytes of seismic data are archived at the Incorporated Research Institutions for Seismology (IRIS) alone. The massive amount of data highlights the need for more efficient and powerful tools for data processing and analyses. The main challenge is the efficient extraction of as much useful information as possible from these large datasets. This is where rapidly evolving machine learning (ML) approaches have the potential to play a key role (Zhu and Beroza 2018; Li et al, 2018; Ross et al, 2018b; Chen 2018). 1 One of the first stages that observational seismologists need to meet this challenge is in the processing of continuous data to detect earthquake signals. Among a large variety of detection methods developed in past few decades, STA/LTA (Allen, 1978) and template matching (Gibbons and Ringdal 2006; Shelly et al. 2007; Ross et al, 2017; Li et al, 2018) are the most commonly used algorithms. While STA/LTA is generalized and efficient, its sensitivity to timevarying background noise and lack of sensitivity to small events, false positives, and events recorded shortly after each other make it less than optimal for robust and sensitive detection. Although the high sensitivity of cross-correlation improves the detection threshold of template matching, the requirement of prior knowledge of templates and multiple cross-correlation procedures make it less general and inefficient for real-time processing of large seismic data volumes. Although more advanced algorithms such as Fingerprint And Similarity Thresholding (FAST) (Yoon et al, 2015) can improve the efficiency of the similarity search, the outputs are in that case limited to repeated events. Shallow Neural Networks (NN) are among the first ML methods used for the earthquake signal detection (e.g.

We present a new method for uncertainty estimation and out-of-distribution detection in neural networks with softmax output. We extend softmax layer with an additional constant input. The corresponding additional output is able to represent the uncertainty of the network. The proposed method requires neither additional parameters nor multiple forward passes nor input preprocessing nor out-of-distribution datasets. We show that our method performs comparably to more computationally expensive methods and outperforms baselines on our experiments from image recognition and sentiment analysis domains. The applications of computational learning systems might cause intrusive effects if we assume that predictions are always as accurate as during the experimental phase. Examples include misclassified traffic signs (Evtimov et al., 2018) and an image tagger that classified two African Americans as gorillas (Curtis, 2015). This is often caused by overconfidence of models that has been observed in the case of deep neural networks (Guo et al., 2017). Such malfunctions can be prevented if we estimate correctly the uncertainty of the machine learning system.

Experience replay (ER) is crucial for attaining high data-efficiency in off-policy reinforcement learning (RL). ER entails the recall of experiences obtained in past iterations to compute gradient estimates for the current policy. However, the accuracy of such updates may deteriorate when the policy diverges from past behaviors, possibly undermining the effectiveness of ER. Previous off-policy RL algorithms mitigated this issue by tuning hyper-parameters in order to abate policy changes. We propose a method for ER that relies on systematically Remembering and Forgetting past behaviors (ReF-ER). ReF-ER forgets experiences that would be too unlikely with the current policy and constrains policy changes within a trust region of the behaviors in the replay memory. We couple ReF-ER with Q-learning, deterministic policy gradient and off-policy gradient methods and we show that ReF-ER reliably improves the performance of continuous-action off-policy RL. We complement ReF-ER with a novel off-policy actor-critic algorithm (RACER) for continuous-action control. RACER employs a computationally efficient closed-form approximation of the action values and is shown to be highly competitive with state-of-the-art algorithms on benchmark problems, while being robust to large hyper-parameter variations.