Deep learning and artificial intelligence are one of the hottest topics in the world today. We see an ever-increasing number of applications that employ deep learning: facial recognition, speech recognition (Siri, 'OK Google', Alexa), Self-Driving Cars, and the list goes on and on. So as a student, fresh employee, team manager, senior management, we get curious: Will this ever-rising wave of AI technology eventually make my job or future career less relevant? That was actually how I stumbled upon the world of Deep Learning years ago, and ended up where I am today: pursuing a postgraduate degree in this field of research. We all have been conditioned to learn things in a certain way: search for information online, sign up for courses, watch tutorials on YouTube. However, the massive and ever-increasing amount of information on this subject is scary, and many of us give up on the pursuit of new, relevant skills quickly.

Since around November 2013, the term'deep learning' started gaining popularity, especially within the data science community. This trend comes shortly after the'big data' boom in 2010 and the'data science' boom in 2011. The upticks in interest are not surprising because companies now realized that they needed individuals who were capable of deciphering insights from the information tsunami that was now present. With data science now being referred to "almost everything that has something to do with data", the process of data utilization evolved beyond data collection and analysis. Now, it was possible to use large sets of data to accurately model events and create data applications.

In the following sections, I will elaborate on how we rapidly built a COVID-19 solution using deep learning tools. The ideas and methods presented here can be used for any new virus or disease with imaging features in CT, especially in the initial phase when data is almost not available. CT scan include a series of slices (for those who are not familiar with CT read short explanation below). Since we had a very limited number of COVID-19 patient's scans, we decided to use 2D slices instead of 3D volume of each scan. This allowed us to multiple our data set and to overcome the first obstacle of a small dataset.

With recent advances, the tech industry is leaving the confines of narrow artificial intelligence (AI) and entering a twilight zone, an ill-defined area between narrow and general AI. To date, all the capabilities attributed to machine learning and AI have been in the category of narrow AI. No matter how sophisticated – from insurance rating to fraud detection to manufacturing quality control and aerial dogfights or even aiding with nuclear fission research – each algorithm has only been able to meet a single purpose. This means a couple of things: 1) an algorithm designed to do one thing (say, identify objects) cannot be used for anything else (play a video game, for example), and 2) anything one algorithm "learns" cannot be effectively transferred to another algorithm designed to fulfill a different specific purpose. For example, AlphaGO, the algorithm that outperformed the human world champion at the game of Go, cannot play other games, despite those games being much simpler.

Machine learning-fuelled artificial intelligence will match and surpass human capacities in the areas of computer vision and speech recognition within five to ten years says – Facebook CEO Mark Zuckerberg. Most of the huge web companies, like Facebook (NASDAQ: FB), utilizes machine learning technology to use its huge data set and deliver better services to its clients. Algorithms work in the background at Facebook to do things like prescribe recommend new connections to Facebook users, to introduce content that matches a user's interest, and to block spam. However, the organization is beginning to utilize machine learning in further advanced ways, for example, for facial recognition on pictures presented on the site. Subsequent to identifying an individual in a photograph, the new Moments application can even recommend to a user that she shares it with that individual.

Graph Neural Networks extend the learning bias imposed by Convolutional Neural Networks and Recurrent Neural Networks by generalising the concept of "proximity", allowing us to have arbitrarily complex connections to handle not only traffic ahead or behind us, but also along adjacent and intersecting roads. In a Graph Neural Network, adjacent nodes pass messages to each other. By keeping this structure, we impose a locality bias where nodes will find it easier to rely on adjacent nodes (this only requires one message passing step). These mechanisms allow Graph Neural Networks to capitalise on the connectivity structure of the road network more effectively. Our experiments have demonstrated gains in predictive power from expanding to include adjacent roads that are not part of the main road.

Dr. Eli David is a leading AI expert specializing in deep learning and evolutionary computation. He is the co-founder of DeepCube. Over the last several years, deep learning has proved to be the key driver of AI advancement and improvements. Drawing from how the human brain operates, deep learning is responsible for advancing AI applications from computer vision to speech recognition to text and data analysis. Deep learning models are trained in research labs using large amounts of training data to demonstrate how the technology could manifest in real-world deployments.

The design of real-world automated control systems that do everything from regulating the temperature of skyscrapers to running the widget-making machine in the widget factory down the street requires expertise in sophisticated physics-based modeling. The need for this modeling expertise increases operational costs and restricts the applicability of automated control to systems in which marginal operational performance improvements lead to huge economic benefits, according to data scientists. With unlimited access to supercomputers and mountains of data, engineers can train artificial intelligence systems such as deep neural networks, a type of machine learning model, to perform automated control. But many people lack access to the necessary computational power to do so, or the ability to generate the amount of data needed to train a controller that has a deep neural network. What's more, these types of deep neural networks are so-called black-box models, which means that the factors they use to make decisions are hidden from the end user.

Despite the large number of patients in Electronic Health Records (EHRs), the subset of usable data for modeling outcomes of specific phenotypes are often imbalanced and of modest size. This can be attributed to the uneven coverage of medical concepts in EHRs. In this paper, we propose OMTL, an Ontology-driven Multi-Task Learning framework, that is designed to overcome such data limitations. The key contribution of our work is the effective use of knowledge from a predefined well-established medical relationship graph (ontology) to construct a novel deep learning network architecture that mirrors this ontology. It can effectively leverage knowledge from a well-established medical relationship graph (ontology) by constructing a deep learning network architecture that mirrors this graph. This enables common representations to be shared across related phenotypes, and was found to improve the learning performance. The proposed OMTL naturally allows for multitask learning of different phenotypes on distinct predictive tasks. These phenotypes are tied together by their semantic distance according to the external medical ontology. Using the publicly available MIMIC-III database, we evaluate OMTL and demonstrate its efficacy on several real patient outcome predictions over state-of-the-art multi-task learning schemes.

Finding general evaluation metrics for unsupervised representation learning techniques is a challenging open research question, which recently has become more and more necessary due to the increasing interest in unsupervised methods. Even though these methods promise beneficial representation characteristics, most approaches currently suffer from the objective function mismatch. This mismatch states that the performance on a desired target task can decrease when the unsupervised pretext task is learned too long - especially when both tasks are ill-posed. In this work, we build upon the widely used linear evaluation protocol and define new general evaluation metrics to quantitatively capture the objective function mismatch and the more generic metrics mismatch. We discuss the usability and stability of our protocols on a variety of pretext and target tasks and study mismatches in a wide range of experiments. Thereby we disclose dependencies of the objective function mismatch across several pretext and target tasks with respect to the pretext model's representation size, target model complexity, pretext and target augmentations as well as pretext and target task types.