Minje Kim, an assistant professor of intelligent systems engineering at the School of Informatics and Computing at IU Bloomington, has received a gift from Intel to pursue a method of lowering the power and computing cost of deep learning processes in artificial intelligence. Intel sought a portfolio of research projects focused on compelling new human-computer interaction advancements that have HCI on the precipice of a breakthrough. As smart devices have become more ubiquitous, advances in deep learning have allowed AI to reach a near-human level. Deep learning allows complicated intelligence jobs -- such as computer vision, near real-time language translation and music recognition to be performed quickly, but such computing comes at a cost. Because neural networks present each of the millions of parameters of a computation in up to 64-bit forms, the computations required are both sizeable and hungry for power.

The superior performance of deep learning relative to other machine learning methodologies has been commented in several forums and magazines in recent times. I would like to post today on three reasons that, in my opinion, are the basis of this commented superiority. I am not the first to comment on this issue [1], and for sure I won't be the last. But I would like to extend the discussion by taking into account the practical reasons behind the success of deep learning. Hence if you are looking for its theoretical background you would do better to look for it in the deep learning literature, where the "Hamiltonian of the spin glass model [2]", the exploitation of compositional functions to cope with the curse of dimensionality [3], their capability to best represent the simplicity of physics-based functions [4], and the flattening of the data manifolds [5] have been proposed.

The role of Artificial Intelligence (AI) as a major catalyst in the healthcare revolution is unquestionable. We are today experiencing the Fourth Industrial Revolution, and the proliferation of technologies that are fusing the physical, digital and biological worlds and thereby impacting global economies and industries is unparalleled. While we are seeing the pervasive influence of technology in our lifestyle, we are challenged by the burden of chronic disease on our healthcare system. In the United States chronic disease accounts for $3 of every $4 spent on healthcare or $7,900 for every American with chronic disease. Chronic disease is both predictable and preventable, and AI can play a pivotal role in addressing solutions that can provide personalized medicine, and interventions.

Neural networks are all the rage in Silicon Valley, infusing so many internet services with so many forms of artificial intelligence. But as good as they may be at recognizing cats in your online photos, AI researchers know that neural networks are still quite flawed, so much so that some wonder whether these pattern recognition systems are a viable path to more advanced--and more reliable--forms of AI. Able to learn tasks by analyzing vast amounts of data, neural networks power everything from face recognition at Facebook to translation at Microsoft to internet search at Google. They're beginning to help chatbots learn the art of conversation. But because they can't make sense of the world without help from such large amounts of carefully labelled data, they aren't suited to everything.

Friday, 3rd of February, 2017, Baltimore, MD - Insilico Medicine today announced that it signed a Memorandum of Understanding (MOU) and started the first collaborative research project with one of the largest research and medical networks, Gachon University and Gil Medical Center. The intent of the long-term collaboration is to develop artificially intelligent multimodal biomarkers of aging and health status as well as interventions intended to slow down or even reverse the processes leading to the age-related loss of function. "We are happy to collaborate with Insilico Medicine, one of the leaders in AI with a specific focus on practical aging research in the pharmaceutical and healthcare industries. The field of artificial intelligence is rapidly evolving and in addition to our own cutting-edge research programs, we collaborate with other leaders to expedite progress and ensure that we can save and extend human life sooner", said Dr. Lee Uhn, Director of AI-based Precision Medicine at Gachon University, Gil Medical Center. The first MOU between the companies was signed on November 18th, but the first project launched and data exchange transpired in January 2017.

Predictive Analytics – We can leverage the power of AI based predictive algorithms to analyze stress and emotion response. This can be used by analyzing data from images via deep learning micro-expression analysis e.g. Google's, Im2Calories leverages deep learning algorithm to analyze food and estimate calories on the plate. Medical imagery is especially amenable to machine-learning. Moorfields Eye Hospital in London announced that it was working with Google's AI research division, DeepMind, to develop an AI system to spot sight-threatening conditions in digital scans of the eye.

One day neural networks could help marketers with their online campaigns. Google's DeepMind is a tantalising glimpse of that future. It can not only recognise and analyse images in a video but in one publicised example, discovered cats. Yes, through its deep learning algorithms, without human guidance, it figured that YouTube is the home of cat videos. It identified 10 million of them (no doubt including my all-time favourite: keyboard cat).

Phrases such as "machine learning" and "artificial intelligence" are thrown around so often by so many people, they risk becoming buzzwords along the lines of "Big Data." But unlike "Big Data," which was always a somewhat-nebulous term, "machine learning" is a definitive process that, when applied correctly, can result in some impressive feats. For instance, check out how Google used it to radically transform the sophistication of Google Translate, one of its core services. For those tech professionals who wish to break into machine learning and artificial intelligence, make no mistake about it: there's a lot of education and training involved. Google wants to make that journey a little easier, though, with a new three-hour course that offers a quick overview of deep-learning fundamentals.

A group of Chinese ophthalmologists and scientists developed a deep learning algorithm to identify congenital cataracts, a rare eye disease responsible for nearly 10 percent of all vision loss in children worldwide. The researchers suggest the algorithm would assist humans, instead of replacing them. "For doctors, technology is not sufficient to determine the best course of treatment with 100 percent certainty, and doctors should therefore make good use of the machine's suggestion to identify and prevent the potential misclassification and complement their own judgment," said study co-author Haotian Lin, a professor of ophthalmology at Sun Yat-sen University. "The results of our comparative analysis showed that both artificial intelligence and human intelligence have strengths and limitations." Using CUDA, TITAN X Pascal GPUs and the cuDNN-accelerated Caffe deep learning framework, the researchers algorithm was able to catch the disease with more than 90% accuracy.

Convolutional Neural Networks (CNN), a technique within the broader Deep Learning field, have been a revolutionary force in Computer Vision applications, especially in the past half-decade or so. One main use-case is that of image classification, e.g. You don't have to limit yourself to a binary classifier of course; CNNs can easily scale to thousands of different classes, as seen in the well-known ImageNet dataset of 1000 classes, used to benchmark computer vision algorithm performance. In the past couple of years, these cutting edge techniques have started to become available to the broader software development community. Industrial strength packages such as Tensorflow have given us the same building blocks that Google uses to write deep learning applications for embedded/mobile devices to scalable clusters in the cloud -- Without having to handcode the GPU matrix operations, partial derivative gradients, and stochastic optimizers that make efficient applications possible.