In any deep learning project, configuring the loss function is one of the most important steps to ensure the model will work in the intended manner. The loss function can give a lot of practical flexibility to your neural networks and it will define how exactly the output of the network is connected with the rest of the network. There are several tasks neural networks can perform, from predicting continuous values like monthly expenditure to classifying discrete classes like cats and dogs. Each different task would require a different type of loss since the output format will be different. For very specialized tasks, it's up to us how we want to define the loss.

AI is poised to have an increasing influence on the way companies create new content, paving the way for new forms of human-machine collaboration. AI is maturing at varying rates around the world, with some organizations using these technologies--including machine learning, deep learning, natural language processing, and computer vision--to support external and internal organizational capabilities. For media and entertainment companies and other content producers in particular, AI may also offer a startling range of possibilities for the creative process, enabling individuals and businesses to generate new content with minimized human input. In a global analysis based on Deloitte's most recent State of AI in the Enterprise survey, early adopters were asked to identify the primary benefits of implementing AI in their organizations.¹ Respondents say using AI to enhance existing products and services is their most sought-after externally focused benefit, with 43% ranking it in their top three, while 31% prioritize using it to optimize external processes. The top internally focused benefit, meanwhile, is optimizing business operations, with 41% placing it among their top three choices, followed by using AI to make better decisions, cited by 34% of respondents.

Since Tesla rolled out the Smart Summon, a feature that enables Model 3 owners to "summon" their car remotely, the technology has been subject to much testing. The limited self-driving technology has seen hundreds of thousands of tests, and many users have published their experience. The results have been mixed, with the mistakes receiving more highlights than the successful performances. Smart Summon is one several technologies that aims to eventually transform Tesla cars into full self-driving vehicles. Tesla activated it for the owners of the Model 3 who had paid a $6,000 upfront fee at the end of September.

"One of the most exciting things about deep learning reconstruction is how it redefines the usual negotiation between exam time and image quality. This could lead to significant downstream value for imaging operations and for patient experience," said Christopher Hess, MD, Chair of the Department of Radiology and Biomedical Imaging at UCSF. SubtleMR delivers a significant improvement in the quality of noisy images, which is particularly beneficial for patients who have difficulty holding still for long periods of time. Artifact-ridden images and the need for re-scans are a challenge for both patients and physicians. SubtleMR integrates seamlessly into the radiology workflow, and it is compatible with any brand of MRI scanner and PACS.

We trained a convolutional neural network (CNN) to classify H.E. stained microscopic images of focal cortical dysplasia type IIb (FCD IIb) and cortical tuber of tuberous sclerosis complex (TSC). Both entities are distinct subtypes of human malformations of cortical development that share histopathological features consisting of neuronal dyslamination with dysmorphic neurons and balloon cells. The microscopic review of routine stainings of such surgical specimens remains challenging. A digital processing pipeline was developed for a series of 56 FCD IIb and TSC cases to obtain 4000 regions of interest and 200.000 sub-samples with different zoom and rotation angles to train a CNN. Our best performing network achieved 91% accuracy and 0.88 AUCROC (area under the receiver operating characteristic curve) on a hold-out test-set.

Researchers from the Allen Institute for AI and University of California, Irvine, have released AllenNLP Interpret, a toolkit for explaining the results from natural-language processing (NLP) models. The extensible toolkit includes several built-in methods for interpretation and visualization components, as well as examples using AllenNLP Interpret to explain the results of state-of-the art NLP models including BERT and RoBERTa. In a paper published on arXiv, the research team described the toolkit in more detail. AllenNLP Interpret uses two gradient-based interpretation methods: saliency maps, which determine how much each word or "token" in the input sentence contributes to the model's prediction, and adversarial attacks, which try to remove or change words in the input while still maintaining the same prediction from the model. These techniques are implemented for a variety of NLP tasks and model architectures.

This makes it perfect for research and production. Over the years it has become one of the most popular libraries for deep learning. The goal of this post is to build an intuition and understanding for how deep learning libraries work under the hood, specifically TensorFlow. To achieve this goal, we will mimic its API and implement its core building blocks from scratch. This has the neat little side effect that, by the end of this post, you will be able to use TensorFlow with confidence, because you'll have a deep conceptual understanding of the inner workings. You will also gain further understanding of things like variables, tensors, sessions or operations. So let's get started, shall we? TensorFlow is a framework composed of two core building blocks -- a library for defining computational graphs and a runtime for executing such graphs on a variety of different hardware. A computational graph has many advantages but more on that in just a moment. Now the question you might ask yourself is, what exactly is a computational graph?

The goal of score following is to track a musical performance, usually in the form of audio, in a corresponding score representation. Established methods mainly rely on computer-readable scores in the form of MIDI or MusicXML and achieve robust and reliable tracking results. Recently, multimodal deep learning methods have been used to follow along musical performances in raw sheet images. Among the current limits of these systems is that they require a non trivial amount of preprocessing steps that unravel the raw sheet image into a single long system of staves. The current work is an attempt at removing this particular limitation. We propose an architecture capable of estimating matching score positions directly within entire unprocessed sheet images. We argue that this is a necessary first step towards a fully integrated score following system that does not rely on any preprocessing steps such as optical music recognition.

Sporting events are extremely complex and require a multitude of metrics to accurate describe the event. When making multiple predictions, one should make them from a single source to keep consistency across the predictions. We present a multi-task learning method of generating multiple predictions for analysis via a single prediction source. To enable this approach, we utilize a fine-grain representation using fine-grain spatial data using a wide-and-deep learning approach. Additionally, our approach can predict distributions rather than single point values. We highlighted the utility of our approach on the sport of Rugby League and call our prediction engine "Rugby-Bot".

We explore the impact of learning paradigms on training deep neural networks for the Travelling Salesman Problem. We design controlled experiments to train supervised learning (SL) and reinforcement learning (RL) models on fixed graph sizes up to 100 nodes, and evaluate them on variable sized graphs up to 500 nodes. Beyond not needing labelled data, out results reveal favorable properties of RL over SL: RL training leads to better emergent generalization to variable graph sizes and is a key component for learning scale-invariant solvers for novel combinatorial problems.