But wait… What is Tensorflow? Tensorflow is a Deep Learning Framework by Google, which released its 2nd version in 2019. It is one of the world's most famous Deep Learning frameworks widely used by Industry Specialists and Researchers. Tensorflow v1 was difficult to use and understand as it was less Pythonic, but with v2 released with Keras now fully synchronized with Tensorflow.keras, it is easy to use, easy to learn, and simple to understand. Remember, this is not a post on Deep Learning so I expect you to be aware of Deep Learning terms and the basic ideas behind it.

CV is a nascent market but it contains a plethora of both big technology companies and disruptors. Technology players with large sets of visual data are leading the pack in CV, with Chinese and US tech giants dominating each segment of the value chain. Google has been at the forefront of CV applications since 2012. Over the years the company has hired several ML experts. In 2014 it acquired the deep learning start-up DeepMind. Google's biggest asset is its wealth of customer data provided by their search business and YouTube.

I've spent the last few months preparing for and applying for data science jobs. It's possible the data science world may reject me and my lack of both experience and a credential above a bachelors degree, in which case I'll do something else. Regardless of what lies in store for my future, I think I've gotten a good grasp of the mindset underlying machine learning and how it differs from traditional statistics, so I thought I'd write about it for those who have a similar background to me considering a similar move.1 This post is geared toward people who are excellent at statistics but don't really "get" machine learning and want to understand the gist of it in about 15 minutes of reading. If you have a traditional academic stats backgrounds (be it econometrics, biostatistics, psychometrics, etc.), there are two good reasons to learn more about data science: The world of data science is, in many ways, hiding in plain sight from the more academically-minded quantitative disciplines.

Is this artificial intelligence or a time machine? Bas Uterwijk, an Amsterdam-based artist, is using AI to create extremely lifelike photographs of historical figures and monuments such as the Statue of Liberty, artist Vincent van Gogh, George Washington and Queen Elizabeth I. Using a program called Artbreeder, which is described as "deep learning software," Uterwijk builds his photographs based on a compilation of portraits, reports the Daily Mail. The program pinpoints common facial features and photograph qualities to produce an image. "I try to guide the software to a credible outcome. I think of my work more as artistic interpretations than scientifically or historically accurate," the artist tells the outlet.

This article is part of Demystifying AI, a series of posts that (try to) disambiguate the jargon and myths surrounding AI. Twenty years ago, the people interested in artificial intelligence research were mostly confined in universities and non-profit AI labs. AI research projects were mostly long-term engagements that spanned across several years--or even decades-- and the goal was to serve science and expand human knowledge. But in the past decade, thanks to advances in deep learning and artificial neural networks, the AI industry has undergone a dramatic change. Today, AI has found its way into many practical applications.

There are four major ways to train deep learning networks: supervised, unsupervised, semi-supervised, and reinforcement learning. We'll explain the intuitions behind each of the these methods. Along the way, we'll share terms you'll read in the literature in parentheses and point to more resources for the mathematically inclined. By the way, these categories span both traditional machine learning algorithms and the newer, fancier deep learning algorithms. For the math-inclined, see this Stanford tutorial which covers supervised and unsupervised learning and includes code samples.

Himax Technologies, Inc., a leading supplier and fabless manufacturer of display drivers and other semiconductor products, announced the launch of WiseEye WE-I Plus HX6537-A solution that supports Google's TensorFlow Lite for Microcontrollers. In this collaboration, Himax is providing the HX6537-A processor with NN (neural network) based SDK (Software Development Kit) for developers to generate deep learning inferences running on TensorFlow Lite for Microcontrollers kernel to boost overall system AI performance. With support to TensorFlow Lite for Microcontrollers, developers are able to take advantage of the WE-I Plus platform as well as the integrated ecosystem from TensorFlow Lite for Microcontrollers to develop their NN based edge AI applications targeted for Notebook, TV, Home Appliance, Battery Camera and IP Surveillance edge computing markets. The benefits of the Himax HX6537-A processor are driven by three unique features. The HX6537-A processor adopts a programmable DSP that runs at 400MHz with power-efficient and multi-level power schemes that incorporate CDM, HOG and JPEG hardware accelerators for real-time motion detection, object detection and image processing.

They may seem tiny and fragile, but mosquitoes can be extremely dangerous. Malaria has been a notoriously life-threatening disease for people of all ages which is spread by mosquitoes. More so because during the initial stages, the symptoms could easily be mistaken for fever, flu, or the common cold. But, in the advanced stages, it could wreak havoc by infecting and rupturing cell structure which could be potentially life-threatening. And if left untreated, it could even result in death.

Humans learn all their life long. They accumulate knowledge from a sequence of learning experiences and remember the essential concepts without forgetting what they have learned previously. Artificial neural networks struggle to learn similarly. They often rely on data rigorously preprocessed to learn solutions to specific problems such as classification or regression. In particular, they forget their past learning experiences if trained on new ones. Therefore, artificial neural networks are often inept to deal with real-life settings such as an autonomous-robot that has to learn on-line to adapt to new situations and overcome new problems without forgetting its past learning-experiences. Continual learning (CL) is a branch of machine learning addressing this type of problem. Continual algorithms are designed to accumulate and improve knowledge in a curriculum of learning-experiences without forgetting. In this thesis, we propose to explore continual algorithms with replay processes. Replay processes gather together rehearsal methods and generative replay methods. Generative Replay consists of regenerating past learning experiences with a generative model to remember them. Rehearsal consists of saving a core-set of samples from past learning experiences to rehearse them later. The replay processes make possible a compromise between optimizing the current learning objective and the past ones enabling learning without forgetting in sequences of tasks settings. We show that they are very promising methods for continual learning. Notably, they enable the re-evaluation of past data with new knowledge and the confrontation of data from different learning-experiences. We demonstrate their ability to learn continually through unsupervised learning, supervised learning and reinforcement learning tasks.

We present an algorithm to learn the relevant latent variables of a large-scale discretized physical system and predict its time evolution using thermodynamically-consistent deep neural networks. Our method relies on sparse autoencoders, which reduce the dimensionality of the full order model to a set of sparse latent variables with no prior knowledge of the coded space dimensionality. Then, a second neural network is trained to learn the metriplectic structure of those reduced physical variables and predict its time evolution with a so-called structure-preserving neural network. This data-based integrator is guaranteed to conserve the total energy of the system and the entropy inequality, and can be applied to both conservative and dissipative systems. The integrated paths can then be decoded to the original full-dimensional manifold and be compared to the ground truth solution. This method is tested with two examples applied to fluid and solid mechanics.