Deep learning models frequently make incorrect predictions with high confidence when presented with test examples that are not well represented in their training dataset. We propose a novel and straightforward approach to estimate prediction uncertainty in a pre-trained neural network model. Our method estimates the training data density in representation space for a novel input. A neural network model then uses this information to determine whether we expect the pre-trained model to make a correct prediction. This uncertainty model is trained by predicting in-distribution errors, but can detect out-of-distribution data without having seen any such example. We test our method for a state-of-the art image classification model in the settings of both in-distribution uncertainty estimation as well as out-of-distribution detection. We compare our method to several baselines and set the state-of-the art for out-of-distribution detection in the Imagenet dataset.

Reinforcement learning can interact with the environment and is suitable for applications in decision control systems. Therefore, we used the reinforcement learning method to establish a foreign exchange transaction, avoiding the long-standing problem of unstable trends in deep learning predictions. In the system design, we optimized the Sure-Fire statistical arbitrage policy, set three different actions, encoded the continuous price over a period of time into a heat-map view of the Gramian Angular Field (GAF) and compared the Deep Q Learning (DQN) and Proximal Policy Optimization (PPO) algorithms. To test feasibility, we analyzed three currency pairs, namely EUR/USD, GBP/USD, and AUD/USD. We trained the data in units of four hours from 1 August 2018 to 30 November 2018 and tested model performance using data between 1 December 2018 and 31 December 2018. The test results of the various models indicated that favorable investment performance was achieved as long as the model was able to handle complex and random processes and the state was able to describe the environment, validating the feasibility of reinforcement learning in the development of trading strategies.

Magic is the art of producing in the spectator an illusion of impossibility. Although the scientific study of magic is in its infancy, the advent of recent tracking algorithms based on deep learning allow now to quantify the skills of the magician in naturalistic conditions at unprecedented resolution and robustness. In this study, we deconstructed stage magic into purely motor maneuvers and trained an artificial neural network (DeepLabCut) to follow coins as a professional magician made them appear and disappear in a series of tricks. Rather than using AI as a mere tracking tool, we conceived it as an "artificial spectator". When the coins were not visible, the algorithm was trained to infer their location as a human spectator would (i.e. in the left fist). This created situations where the human was fooled while AI (as seen by a human) was not, and vice versa. Magic from the perspective of the machine reveals our own cognitive biases.

Purpose: Manual feedback from senior surgeons observing less experienced trainees is a laborious task that is very expensive, time-consuming and prone to subjectivity. With the number of surgical procedures increasing annually, there is an unprecedented need to provide an accurate, objective and automatic evaluation of trainees' surgical skills in order to improve surgical practice. Methods: In this paper, we designed a convolutional neural network (CNN) to classify surgical skills by extracting latent patterns in the trainees' motions performed during robotic surgery. The method is validated on the JIGSAWS dataset for two surgical skills evaluation tasks: classification and regression. Results: Our results show that deep neural networks constitute robust machine learning models that are able to reach new competitive state-of-the-art performance on the JIGSAWS dataset. While we leveraged from CNNs' efficiency, we were able to minimize its black-box effect using the class activation map technique. Conclusions: This characteristic allowed our method to automatically pinpoint which parts of the surgery influenced the skill evaluation the most, thus allowing us to explain a surgical skill classification and provide surgeons with a novel personalized feedback technique. We believe this type of interpretable machine learning model could integrate within "Operation Room 2.0" and support novice surgeons in improving their skills to eventually become experts.

Facial recognition software could be used to detect hail storms - and their severity. That's according to scientists at the US National Center for Atmospheric Research, who've tested the software's effectiveness on meteorological data. Specifically, they found that a deep learning model called a convolutional neural network can spot the early signs as they happen - better than current methods. The promising results, published in the American Meteorological Society's Monthly Weather Review, could be a game-changer for providing accurate weather warnings. AI: The promising results, published in the American Meteorological Society's Monthly Weather Review, could be a game-changer for providing accurate weather warning Whether or not a storm produces hail hinges on myriad meteorological factors.

HOSTKEY deploys a well-established environment for machine learning applications such as neural networks with high-performance GPUs and dedicated servers with NVIDIA GTX 1080/1080Ti and RTX 2080Ti graphics cards. Just start your TensorFlow experience in a straightforward and user-friendly environment making it easy to build, train and deploy machine learning models at scale. TensorFlow runs up to 50% faster on our high-performance GPUs and scales easily. Now your machines learn in hours, not days. Deep Learning is a buzzword that will be familiar to most people.

The first video shows the set-up and process of optimizing a simple parametric design for an neighborhood. The algorithm takes five fitness objectives into account: Solar comfort on the streets (weighted by pedestrian frequency); wind comfort; footfall through the neighborhood; access to the neighborhood, overall access to local transit stations; The latter two indicators are computed for the whole area, thus enabling to include a positive impact of the new quarter's spatial arrangement on the whole neighborhood as a goal dimension. By using our deep learning based predictions for solar and wind related measures, one iteration takes just about three seconds to be computed.

There are many options to do the course work, e.g., AWS, PaperSpace, etc., but I found Google Colaboratory is the best and easiest option. Here is the instruction for the Fast.ai Unlike other option, Colab guarantees to work because Google starts with a clean, new virtual machine (VM) every time, and in the first few steps in the notebooks, it loads the required correct version of Pytorch and Fast.ai.

AI, or Artificial Intelligence, is often demonised and portrayed as some cyborg entity just about ready to take our jobs and eventually kill us all, but more and more businesses, martech and adtech providers are using different AI subsystems each day to advance their services. The term AI is contentiously used to describe a broad spectrum of systems and software's, the controversy arises from where we can begin to describe a machine as being'intelligent' opposed to simply following complex but nonetheless human-reliant algorithms. Regardless of strict definition, there are helpful systems within the subsets of AI which already exist that B2B marketers need to utilise. Machine learning is a subset of AI that can help marketers to improve productivity by taking over mundane tasks, particularly work involving dissecting datasets (like our Argus platform for example). If you're not already using some forms of machine learning, it might be helpful to understand why some sytstems have been reported to increase the productivity of business by 40% (Source: Accenture) and how you can effectively incorporate machine learning into your marketing strategy.

Have you ever wondered whom to thank for some of the modern conveniences you might have started taking for granted, like Siri, Cortana or Alexa (assuming you agree these are conveniences)? The people at the Association for Computing Machinery (ACM) decided to thank Geoffrey Hinton, Yoshua Bengio and Yann LeCun in April of this year by honoring them with the Turing Award for their contributions to deep learning and neural networks. These contributions are put to use every time you log into your smartphone using fingerprint or facial recognition or when you use Google Photos or a voice assistant, and likely every time you use Amazon, Netflix, Facebook or Instagram. The advances in automatic language translation and autonomous cars in recent years arguably wouldn't have progressed as rapidly had it not been for the contributions of these three researchers. All of that is still an understatement of their contributions to artificial intelligence (AI).