Although technology has come a long way in recent years, not least in terms of the immense power and resources available through cloud computing services let alone the vast amount of data that can be allocated to cloud storage, computers and machines still can't match the power of the human brain. What makes humans so unique is that we can learn as we go, drawing on our memories and experiences. That means taking in data from the world around us and forming ideas about how to optimally perform tasks or understand new information. Deep learning, which is a branch of artificial intelligence, aims to replicate our ability to learn and evolve in machines. At the end of the day, deep learning allows computers to take in new information, decipher it, and produce an output--all without humans needing to be involved in the process.

Despite its perceived transformational potential, deep learning remains a machine learning'holy grail' for the majority of AI specialists. That's the conclusion of research from Peltarion, based on the responses of 350 AI decision-makers from the UK and the Nordics with direct responsibility for "shepherding AI" at companies with more than 1000 employees. The resulting report sought to explore the understanding of deep learning versus other types of machine learning practices, and the barriers holding businesses back from taking it from an ideal to reality. What the findings seemed to show is that while businesses are appointing individuals to oversee AI, and there is indeed confidence in the technology's potential, there is still a clear knowledge gap in the workings of the tech-- even among the so-called specialists who make the decisions to invest and develop it. According to the report, 99 percent of respondents thought that deep learning would transform their industry-- 32 percent said it would "totally" transform it. At the same time, just 60 percent were confident about what deep learning is and how it works, and just 1 percent had deployed it extensively.

Machine Intelligence brings a potent combination of AI and IoT to businesses. This domain includes industrial automation, application of deep learning to sensor telemetry, edge computing, intelligent video analytics and cognitive computing. First-generation IoT platforms and solutions relied on a basic rules engine to orchestrate the workflow among the connected sensors and actuators. These rules engines were comparable to the functionality provided by IFTTT. With AIoT, the complex event processing engines and stream analytics are natively integrated with AI.

Amazon's DJL is a deep learning toolkit used to develop machine learning (ML) and deep learning (DL) models natively in Java while simplifying the use of deep learning frameworks. A toolkit open-sourced just in time for re:Invent 2019, DJL provides a set of high-level APIs to train, test and run inference. Java developers can develop their own models or utilize pre-trained models developed by data scientists in Python from their Java code. DJL stays true to Java's motto, "write once, run anywhere (WORA)", by being engine and deep learning framework-agnostic. Developers can write code once that runs on any engine.

Almost a third (32%) said that deep learning would'totally' transform their industry, while 26% said other kinds of machine learning would have this effect. Most CTOs highlight machine learning as the technology that will disrupt their industry and lead to new innovations. But, how can they effectively put machine learning models into production? However, only 60% said they were confident that they knew what deep learning is and how it works, despite participants having direct responsibility for overseeing AI. The sample was made up of 350 CIOs and senior AI decision makers in the UK and the Nordics, all of whom worked for organisations of at least 1,000 employees.

NVIDIA today introduced groundbreaking inference software that developers everywhere can use to deliver conversational AI applications, slashing inference latency that until now has impeded true, interactive engagement. NVIDIA TensorRT 7 -- the seventh generation of the company's inference software development kit -- opens the door to smarter human-to-AI interactions, enabling real-time engagement with applications such as voice agents, chatbots and recommendation engines. It is estimated that there are 3.25 billion digital voice assistants being used in devices around the world, according to Juniper Research. By 2023, that number is expected to reach 8 billion, more than the world's total population. TensorRT 7 features a new deep learning compiler designed to automatically optimize and accelerate the increasingly complex recurrent and transformer-based neural networks needed for AI speech applications.

Google-owned DeepMind is to collaborate with a UK hospital to help doctors spot more than 50 different eye diseases using AI. When it isn't creating artificial intelligence (AI) capable of destroying human opponents in a game of Go, Google-owned DeepMind is trying to build other systems that could transform healthcare, among other things. Now, the UK-based company has revealed a joint research partnership with Moorfields Eye Hospital that could help spot sight-threatening eye diseases much quicker than before. Publishing its findings in Nature Medicine, the company said that its latest AI can quickly run through eye scans taken from routine clinical practice and identify more than 50 serious diseases as accurately as world-leading expert doctors. Under existing systems, ophthalmologists use 3D images called optical coherence tomography (OCT) to create a detailed map of a person's eye.

Hackathons are a great way to learn new skills and build your personal brand. They push you to put your creative hats on and apply your theoretical and practical knowledge in an innovative manner. Generally, the problems given at Hackathons are not always very straight forward and require a good amount of understanding in the subject. This becomes quite difficult if you are a working professional to manage time and shift your focus in between contexts. I participated in Analytics Vidhya's Game of Deep Learning Hackathon.

Multimodal analysis that uses numerical time series and textual corpora as input data sources is becoming a promising approach, especially in the financial industry. However, the main focus of such analysis has been on achieving high prediction accuracy while little effort has been spent on the important task of understanding the association between the two data modalities. Performance on the time series hence receives little explanation though human-understandable textual information is available. In this work, we address the problem of given a numerical time series, and a general corpus of textual stories collected in the same period of the time series, the task is to timely discover a succinct set of textual stories associated with that time series. Towards this goal, we propose a novel multi-modal neural model called MSIN that jointly learns both numerical time series and categorical text articles in order to unearth the association between them. Through multiple steps of data interrelation between the two data modalities, MSIN learns to focus on a small subset of text articles that best align with the performance in the time series. This succinct set is timely discovered and presented as recommended documents, acting as automated information filtering, for the given time series. We empirically evaluate the performance of our model on discovering relevant news articles for two stock time series from Apple and Google companies, along with the daily news articles collected from the Thomson Reuters over a period of seven consecutive years. The experimental results demonstrate that MSIN achieves up to 84.9% and 87.2% in recalling the ground truth articles respectively to the two examined time series, far more superior to state-of-the-art algorithms that rely on conventional attention mechanism in deep learning.

The success of deep neural networks in many real-world applications is leading to new challenges in building more efficient architectures. One effective way of making networks more efficient is neural network compression. We provide an overview of existing neural network compression methods that can be used to make neural networks more efficient by changing the architecture of the network. First, we introduce a new way to categorize all published compression methods, based on the amount of data and compute needed to make the methods work in practice. These are three 'levels of compression solutions'. Second, we provide a taxonomy of tensor factorization based and probabilistic compression methods. Finally, we perform an extensive evaluation of different compression techniques from the literature for models trained on ImageNet. We show that SVD and probabilistic compression or pruning methods are complementary and give the best results of all the considered methods. We also provide practical ways to combine them.