Montreal-based start-up Lyrebird is looking to change that with an artificially intelligent system that learns to mimic a person's voice by analyzing speech recordings and the corresponding text transcripts as well as identifying the relationships between them. Introduced last week, Lyrebird's speech synthesis can generate thousands of sentences per second--significantly faster than existing methods--and mimic just about any voice, an advancement that raises ethical questions about how the technology might be used and misused. The ability to generate natural-sounding speech has long been a core challenge for computer programs that transform text into spoken words. Artificial intelligence (AI) personal assistants such as Siri, Alexa, Microsoft's Cortana and the Google Assistant all use text-to-speech software to create a more convenient interface with their users. Those systems work by cobbling together words and phrases from prerecorded files of one particular voice.

Last week I gave a presentation around Artificial Intelligence to Product people: PO's, PMO's and others. I recalled how my geek day is using Alexa, Slack bots, x.ai, Smart Lighting & how happy I am of in living in such an era. Many of the smart, cognitive and automation things that we see are simple workflows, and they do not have AI, per se, on it (a discussion for another post). In a great conversation I tried to pass my message along, elucidate some concepts and show that Machine Learning algorithms can be used in lots of products -- so, you can have AI in the backend of your app, software, API, website, … Speech recognition, recommendation systems, image tagging, NLP, search optimisation, clustering and many others features that your product can react to it, providing a personalised experience. Telling the history of AI, the overselling of the 50's and 60's, the AI Winter and the recent boom I try to make everyone cautious about overselling its capabilities.

Today, read any tech article or news and you will be fired with the terms "Artificial Intelligence", "Machine Learning" and "Deep Learning". The biggest corporate giants Google, IBM, Facebook, Microsoft and Amazon are voraciously acquiring Artificial Intelligence startups and companies. In just 3 months of 2017, 34 acquisitions were made. Forrester in the new report, "Prediction 2017: Artificial Intelligence Will Drive the Insights Revolution", predicts a 300% increase in investment in Artificial Intelligence from 2016 to 2017. The report further proceeds to say that "insight-driven businesses will steal $ 1.2 trillion per annum from their less informed peer by 2020."

In Hollywood, artificial intelligence is typically associated with dystopian futures in which the machines have become sentient and assumed control. I'm not saying that doomsday scenario is out of the question – I'm many things but I'm not a futurologist. I think we'll learn to coexist just fine with our ultra-smart artificial intelligence overlords, even as we become increasingly reliant on their capacity for deep learning and neural networks. This month I've been delving deep into the matter of AI on behalf of Sparklane, and it's at their site where my detailed blog post on the matter can be found. Venture there to read 5 Ways in Which AI is Transforming B2B Marketing in just a moment, but first here's a quick primer on where we're currently at with artificial intelligence.

The time when companies must face up to the new digital realities that emerging technologies are not only approaching fast but looking to disrupt in the most legacy of industries. Established enterprises, private-held businesses and a rising number of venture-capital-backed startups across the world are moving rapidly to connect their products and equipment to the Internet of Things (IoT), opening up opportunities to create new business models and transform how they run their operations and engage with customers. However, tapping into the IoT is only part of the story. For companies to realise the full potential of IoT enablement, they need to combine IoT with rapidly-advancing Artificial Intelligence (Machine Learning, Deep Learning) technologies, which enable'smart machines' to simulate intelligent behaviour and make well-informed decisions with little or no human intervention. Over the coming years, these ongoing advances in AI will have profound impacts on jobs, skills and HR strategies in virtually every industry--underlining the fact that companies don't have the luxury of time as they map out their plans for an AI-enabled world.

It's just not practical to program a car to drive itself in every environment, given the nearly infinite range of possible variables involved. But, thanks to AI, we can show it how to drive. And, unlike your teenager, you can then see what it's paying attention to. With NVIDIA PilotNet, we created a neural-network-based system that learns to steer a car by observing what people do. We developed a method for the network to tell us what it prioritized when making driving decisions.

China has overtaken the United States to become the world leader in deep learning research, a branch of artificial intelligence (AI) inspired by the human brain. Deep learning algorithms are modeled on biological neural networks and enable machines to learn and mimic human-like responses. Think: smartphone assistants answering your questions, or Amazon recommending products based on your search and purchasing history.

Many companies today are employing deep learning techniques in different facets of their business. Yelp uses deep learning algorithms to feature the best user photos, Netflix uses it to suggest movies you might be interested in, and Google ultimately transformed the concept of deep learning by creating a system that helps generate responses to search queries. It's widely believed that you no longer need to know your data; you can just apply a little deep learning magic and poof -- problem solved. However, the reality is that this could not be further from the truth, at least for the legal industry. As an example, deep learning can be essential when legal counsel within an organization wants to find out how many contracts (among 10s to 100s of thousands) has termination for convenience clauses that could disrupt the business, or if any have strict assignment rules that may be a problem for a M&A event.

This work studies the entity-wise topical behavior from massive network logs. Both the temporal and the spatial relationships of the behavior are explored with the learning architectures combing the recurrent neural network (RNN) and the convolutional neural network (CNN). To make the behavioral data appropriate for the spatial learning in CNN, several reduction steps are taken to form the topical metrics and place them homogeneously like pixels in the images. The experimental result shows both the temporal- and the spatial- gains when compared to a multilayer perceptron (MLP) network. A new learning framework called spatially connected convolutional networks (SCCN) is introduced to more efficiently predict the behavior.

Deep learning has become a powerful and popular tool for a variety of machine learning tasks. However, it is extremely challenging to understand the mechanism of deep learning from a theoretical perspective. In this work, we study robustness of a deep network in its generalization capability against removal of a certain number of connections between layers. A critical value of this number is observed to separate a robust (redundant) regime from a sensitive regime. This empirical behavior is captured qualitatively by a random active path model, where the path from input to output is randomly and independently constructed. The empirical critical value corresponds to termination of a paramagnetic phase in the random active path model. Furthermore, this model provides us qualitative understandings about dropconnect probability commonly used in the dropconnect algorithm and its relationship with the redundancy phenomenon. In addition, we combine the dropconnect and the random feedback alignment for feedforward and backward pass in a deep network training respectively, and observe fast learning and improved test performance in classifying a benchmark handwritten digits dataset.