More than a quarter of a million developers, researchers, innovators, and creators are gearing up for the long-awaited #1 AI conference – NVIDIA GTC which is kick-starting on November 8, 2021. The four-day virtual event will highlight some of the latest advancements in AI, deep learning, data science, high-performance computing (HPC), robotics, data science, networking, graphics and more. The Keynote by Jensen Huang, NVIDIA Founder, President and CEO, named as one of the world's most influential people of 2021, is expected to inspire and showcase the latest developments in AI, new solutions and latest products that will help solve the world's toughest challenges. Don't miss this Keynote, which will be live on November 9, at 1:30 PM IST GTC will provide a great opportunity for developers to learn the advancements in the latest technologies from the world's top innovators, scientists, and researchers. In addition, startups, academia, and the largest enterprises will all come together at GTC, giving participants a unique opportunity to share ideas and collaborate on creating the future.

This High-Intensity 14.5 Hour Bundle Will Help You Help Computers Address Some of Humanity's Biggest Problems In this course, intended to expand upon your knowledge of neural networks and deep learning, you'll harness these concepts for computer vision using convolutional neural networks. Going in-depth on the concept of convolution, you'll discover its wide range of applications, from generating image effects to modeling artificial organs. In this course, you'll dig deep into deep learning, discussing principal components analysis and a popular nonlinear dimensionality reduction technique known as t-distributed stochastic neighbor embedding (t-SNE). From there you'll learn about a special type of unsupervised neural network called the autoencoder, understanding how to link many together to get a better performance out of deep neural networks. A recurrent neural network is a class of artificial neural network where connections form a directed cycle, using their internal memory to process arbitrary sequences of inputs.

Google's DeepMind can predict almost exactly where and when it's going to rain, but for years, the artificial intelligence company couldn't figure out how to stop drowning in debt. On Tuesday, however, DeepMind said it has now actually eked out a profit. Finally: hope for all the plucky little start-ups with the backing of a $1.7 trillion parent company. DeepMind, which is likely the largest AI research operation in the world, has a very unique revenue model: it makes 100% of its money selling the technologies it develops to other subsidiaries of Alphabet, Google's parent company. But while Alphabet brought in $182.5 billion in revenue last year, DeepMind was hemorrhaging money for years -- losing more in both 2018 and 2019 (about $680 million each year) than Google paid for it in 2014 ($545 million). Alphabet also wrote off $1.5 billion in DeepMind's debt in 2019.

All the pieces are coming together: big data, advances in hardware, emerging powerful AI algorithms, and an open source community for tools that reduces barriers to entry for industry and start-ups alike. The result: AI is being propelled out of research labs and into our everyday lives, from navigating cities, ride shares, our energy networks, to the online world. In 2018 everyone is starting to see the business value of AI. It is being added to more and more things every year, and it is getting smarter and smarter – accelerating human innovation. But as AI becomes more powerful, more autonomous and broader in its use and impact, the unsolved issue of AI safety is paramount. Risks include: bias, poor decision making, low transparency, job losses and malevolent use of AI, such as autonomous weaponry.

Researchers led by the Institute of Scientific and Industrial Research (SANKEN) at Osaka University have trained a deep neural network to correctly determine the output state of quantum bits, despite environmental noise. The team's novel approach may allow quantum computers to become much more widely used. Modern computers are based on binary logic, in which each bit is constrained to be either a 1 or a 0. But thanks to the weird rules of quantum mechanics, new experimental systems can achieve increased computing power by allowing quantum bits, also called qubits, to be in "superpositions" of 1 and 0. For example, the spins of electrons confined to tiny islands called quantum dots can be oriented both up and down simultaneously. However, when the final state of a bit is read out, it reverts to the classical behavior of being one orientation or the other. To make quantum computing reliable enough for consumer use, new systems will need to be created that can accurately record the output of each qubit even if there is a lot of noise in the signal.

If you have trouble reading this email, see it on a web browser. I hope you are well. In this issue, we dive into the maturation of the artificial intelligence (AI) ecosystem and its landscape for 2021, updated and curated resources for MLOps, how deep learning is accurately predicting traffic crashes before they occur, a complete data science cheat sheet with resources for stats, probability, ML and more, and some updates from our end. This issue is brought to you thanks to expert.ai: Learn how to apply natural language to one of the most challenging and real-world problems.

In recent years, a wide spectrum of deep learning visualization methods has emerged. For a comprehensive overview of visual analytics (VA) for deep learning, we refer the reader to a survey by Hohman et al. A case study using a VA system to assess a model's performance to detect and classify traffic lights has shown that interactive VA systems can successfully guide experts to improve their training data While these examples all focus on the inspection of a single model, others support model comparisons. For example, using REMAP [cashman_ablate_2020], users can rapidly create model architectures through ablations (i.e., removing single layers of an existing model) and variations (i.e., creating new models through layer replacements) and compare the created models by their structure and performance. Interactive "playgrounds" require relatively little underlying deep learning knowledge and can be used for educational purposes.

Deep learning is used to classify data into several groups based on nonlinear curved surfaces. In this paper, we focus on the theoretical analysis of deep learning using the rectified linear unit (ReLU) activation function. Because layers approximate a nonlinear curved surface, increasing the number of layers improves the approximation accuracy of the curved surface. While neurons perform a layer-by-layer approximation of the most appropriate hyperplanes, increasing their number cannot improve the results obtained via canonical correlation analysis (CCA). These results illustrate the functions of layers and neurons in deep learning with ReLU.

Artificial intelligence and machine learning pioneers are rapidly expanding on techniques that were originally designed for natural language processing and translation to other domains, including critical infrastructure and the genetic language of life. This was reported in the 2021 edition of the State of AI Report by investors Nathan Benaich of Air Street Capital and Ian Hogarth, an angel investor. Started in 2018, their report aims to be a comprehensive survey of trends in research, talent, industry, and politics, with predictions mixed in. The authors are tracking "182 active AI unicorns totaling $1.3 trillion of combined enterprise value" and estimate that exits by AI companies have created $2.3 trillion in enterprise value since 2010. One of their 2020 predictions was that we would see the attention-based transformers architecture for machine learning models branch out from natural language processing to computer vision applications.

Deep-learning techniques have been successfully used for time-series forecasting and have often shown superior performance on many standard benchmark datasets as compared to traditional techniques. Here we present a comprehensive and comparative study of performance of deep-learning techniques for forecasting prices in financial markets. We benchmark state-of-the-art deep-learning baselines, such as NBeats, etc., on data from currency as well as stock markets. We also generate synthetic data using a fuzzy-logic based model of demand driven by technical rules such as moving averages, which are often used by traders. We benchmark the baseline techniques on this synthetic data as well as use it for data augmentation. We also apply gradient-based meta-learning to account for non-stationarity of financial time-series. Our extensive experiments notwithstanding, the surprising result is that the standard ARIMA models outperforms deep-learning even using data augmentation or meta-learning. We conclude by speculating as to why this might be the case.