After breaking all the records related to training computer vision models, NVIDIA now claims that it's AI platform is able to train a natural language neural network model based on one of the largest datasets in a record time. It also claims that the inference time is just 2 milliseconds which translates to an extremely fast response from the model participating in a conversation with a user. After computer vision, natural language processing is one of the top applications of AI. From Siri to Alexa to Cortana to Google Assistant, all conversational user experiences are powered by AI. The advancements in AI research is putting the power of language understanding and conversational interface into the hands of developers.

The majority of NHS users would rather discuss their hospital and GP appointments with an artificial intelligence service instead of a human. That's according to new research from technology startup EBO.ai, which explores the growing role digital communication tools play in the NHS. It found that more than three quarters (76%) of NHS users would be happy to receive automated reminders from an AI-powered virtual assistant, compared to the 58% who'd rather interact with a human. Figures from NHS England claim that missed appointments cost the health service over £216 million annually, and many people believe that AI technology could help minimize this cost. Dr. Gege Gatt, CEO of EBO.ai, said: "The NHS has already invested millions in the latest technologies, but the enormous potential of AI remains largely untapped. "The adoption we have seen thus far has focused on primary patient care including assessment and diagnosis, but AI can improve patient experiences outside the treatment room too." "Virtual assistants help patients manage their care 24/7, with no need to wait for opening hours or spend time on hold in a phone queue.

Oh, how the blind faith in technology keeps spinning its subpar centrifuge until the excessive speed of make-believe catapults its favorite toys into reality. It pays to listen to me, I tell you. I have made similar predictions about the subpriming of venture capital, the deplorable stance of Facebook, the re-risking of asset management, the voodoo of economics, all coming true after the pageantry of positivity is forced to meet reality. We must build more advanced operating-systems for humanity and hold the promises of policy, capital, and innovation to account. I wrote in 2011 how the enormous amount of false-positives would make Siri useless, now 8-years later Apple's former long-time Siri head confirmed my views, Siri does not deliver.

As Apple prepares to launch a new iPhone, Alex Hern explores the privacy scandal around its automated personal assistant, Siri.

The majority of NHS users would rather discuss their hospital and GP appointments with an artificial intelligence service instead of a human. That's according to new research from technology startup EBO.ai, which explores the growing role digital communication tools play in the NHS. It found that more than three quarters (76%) of NHS users would be happy to receive automated reminders from an AI-powered virtual assistant, compared to the 58% who'd rather interact with a human. Figures from NHS England claim that missed appointments cost the health service over £216 annually, and many people believe that AI technology could help minimize this cost. Dr. Gege Gatt, CEO of EBO.ai, said: "The NHS has already invested millions in the latest technologies, but the enormous potential of AI remains largely untapped. "The adoption we have seen thus far has focused on primary patient care including assessment and diagnosis, but AI can improve patient experiences outside the treatment room too." "Virtual assistants help patients manage their care 24/7, with no need to wait for opening hours or spend time on hold in a phone queue.

Using deep learning and artificial intelligence (AI), FAU scientists are bringing to life one of about a handful of these quadruped robots in the world. Astro is unique because he is the only one of these robots with a head, 3D printed to resemble a Doberman pinscher, that contains a (computerized) brain. What would you get if you combined Apple's Siri and Amazon's Alexa with Boston Dynamic's quadruped robots? You'd get "Astro," the four-legged seeing and hearing intelligent robodog. Using deep learning and artificial intelligence (AI), scientists from Florida Atlantic University's Machine Perception and Cognitive Robotics Laboratory (MPCR) in the Center for Complex Systems and Brain Sciences in FAU's Charles E. Schmidt College of Science are bringing to life one of about a handful of these quadruped robots in the world.

Written by Dr. Christine Izuakor for Veriato, a cybersecurity company Artificial intelligence (AI) is used all around us and if you've used some sort of voice activated technology to make your life easier, then there was likely some element of AI involved. Some of the most notable examples include Siri, Amazon Alexa, Google Assistant and Tesla semi-autonomous vehicles. Individual consumers no longer have to fumble around in the dark to flip the light switch at home, manually search playlists for songs, or type in a password to get into smartphones. Similarly, businesses can now analyze millions of data records and find trends that can help them predict things like when their assets may require maintenance or what purchasing decisions customers are likely to make. Thanks to AI, there are automation and optimization solutions for almost everything – including some of our most significant technical challenges.

Combinatorial linear semi-bandits (CLS) are widely applicable frameworks of sequential decision-making, in which a learner chooses a subset of arms from a given set of arms associated with feature vectors. Existing algorithms work poorly for the clustered case, in which the feature vectors form several large clusters. This shortcoming is critical in practice because it can be found in many applications, including recommender systems. In this paper, we clarify why such a shortcoming occurs, and we introduce a key technique of arm-wise randomization to overcome it. We propose two algorithms with this technique: the perturbed C${}^2$UCB (PC${}^2$UCB) and the Thompson sampling (TS). Our empirical evaluation with artificial and real-world datasets demonstrates that the proposed algorithms with the arm-wise randomization technique outperform the existing algorithms without this technique, especially for the clustered case. Our contributions also include theoretical analyses that provide high probability asymptotic regret bounds for our algorithms.

Self-attentional models are a new paradigm for sequence modelling tasks which differ from common sequence modelling methods, such as recurrence-based and convolution-based sequence learning, in the way that their architecture is only based on the attention mechanism. Self-attentional models have been used in the creation of the state-of-the-art models in many NLP tasks such as neural machine translation, but their usage has not been explored for the task of training end-to- end task-oriented dialogue generation systems yet. In this study, we apply these models on the three different datasets for training task-oriented chatbots. Our finding shows that self-attentional models can be exploited to create end-to-end task-oriented chatbots which not only achieve higher evaluation scores compared to recurrence-based models, but also do so more efficiently.

We present Distributed Equivalent Substitution (DES) training, a novel distributed training framework for recommender systems with large-scale dynamic sparse features. Our framework achieves faster convergence with less communication overhead and better computing resource utilization. DES strategy splits a weights-rich operator into sub-operators with co-located weights and aggregates partial results with much smaller communication cost to form a computationally equivalent substitution to the original operator. We show that for different types of models that recommender systems use, we can always find computational equivalent substitutions and splitting strategies for their weights-rich operators with theoretical communication load reduced ranging from 72.26% to 99.77%. We also present an implementation of DES that outperforms state-of-the-art recommender systems. Experiments show that our framework achieves up to 83% communication savings compared to other recommender systems, and can bring up to 4.5x improvement on throughput for deep models.