The last time you heard a computer convert a line of text to speech, it probably jarred. Google's machine-learning division, DeepMind, has developed a new voice synthesis system using artificial intelligence that it thinks will improve the situation. Having a computer generate the sound of a voice isn't a new idea. Perhaps the most common approach is simply to use an incredibly large selection of pre-recorded speech fragments from a single person. In a technique called concatenative synthesis, these are pieced together to create larger sounds, words, and sentences.

There is a massive amount of hype that surrounds AI. How much is fact, how much is fiction? This event is to decipher the status of AI technology currently by providing a group of AI experts to present and chat on the subject. The event will begin with two introduction presentations, one by Azeem Azhar on "Why the Boom Now?" and the second by Christoph Auer-Welsbach on IBM Watson and City.ai. Following, a panel discussion moderated by Tak Lo will ensue on "The Status of AI in Asia Today" with Sinuhe Arroyo, Jason Chiu, and Jeffrey Broer.

Steam rolled out of the pot, and my kitchen was filled with an intoxicating blend of scents. I'd cooked a turkey ragu before, but as soon as I opened the pot, I knew this would be an entirely different experience. There were the familiar, comforting notes of ground turkey and mushrooms, but there was also earthiness from the beets, the fire of serrano chile, and an unexpected brightness from the coriander seed (coriander seed!). My dinner wouldn't have been out of place at one of San Francisco's trendy food trucks or upscale restaurants, but coming out of my 15 IKEA pot, it was unbelievable. As I mixed in the pasta and ricotta cheese and took the first incredible bite, I realized that A.I. could do a lot more for me than just beating me at Go.

I jus' bin dahn the ol' rub-a-dub for a pokey alright sparrah do us a lemon John. Be comforted that I have neither succumbed to Dickvandykitis nor do I have any compulsion to "do the old bamboo" – a suspicious euphemism if ever I heard one. I am merely rejoicing in being told that I am 100% cockney, having successfully completed a painstakingly comprehensive study by the world authority known as The Internet, testing my cultural facility to speak incomprehensible bollocks. I am particularly pleased in being recognised as a genuine Lahndahn cockney since I'm from Yorkshire. Proper Lundunners who I know took the test came out with miserable results compared to mine.

"The second edition was reviewed in JASA by Maiti (2004) … we now stand 10 years later with an even more impressive textbook that truly stands for what Bayesian data analysis should be. Quite a lot … this is truly the reference book for a graduate course on Bayesian statistics and not only Bayesian data analysis." Praise for the Second Edition: … it is simply the best all-around modern book focused on data analysis currently available. The second edition makes this an even more robust choice. I am thoroughly excited to have this book in hand to supplement course material and to offer research collaborators and clients at our consulting lab more sophisticated methods to solve their research problems.

Cancer is a tricky disease to treat. With more than one hundred known types, each responding differently to treatment depending on the person they're growing inside (and dozens of other factors), oncologists certainly have their work cut out for them. Machine learning could soon make their jobs a bit easier. IBM's supercomputer Watson has been applying machine learning to personalized cancer treatment for some time. After pouring over 600,000 medical reports and 1.5 million anonymized patient records and clinical trials, the data should help define the clearest path forward for doctors.

Digital art is sort of a vague term: GIFs? Art in which digital technology is central to the creative process. Some would say necessarily so. Computers--and the technologies we've built with them--have opened up huge new realms of possibility. Machine intelligence (think IBM's Watson or Google Brain) is at the forefront of these new discoveries.

Let's go through a high-level exploration of the evolution of computational hardware technologies with a focus on applications to machine learning (ML), and using cryptocurrency mining as an analogy. I posit that the machine learning industry is undergoing the same progression of hardware as cryptocurrency did years ago. Machine learning algorithms often consist of matrix (and tensor) operations. These calculations benefit greatly from parallel computing, which leads to model-training performed on graphics cards (rather than only on the CPU). Each step in this progression of technologies produces tremendous performance advantages.

Both questions found natural answers in the field of Recurrent Neural Networks (RNNs), which are brain-inspired general purpose computers that can learn parallel-sequential programs or algorithms encoded as weight matrices. Since then, a lot has happened. Some of the most successful applications in machine learning (including deep learning) are now driven by RNNs such as Long Short-Term Memory, e.g., speech recognition, video recognition, natural language processing, image captioning, time series prediction, etc. Reinforcement-learning and evolutionary RNNs are solving complex control tasks from raw video input. Many RNN-based methods learn sequential attention strategies. At this symposium, we will review the latest developments in all of these fields, and focus not only on RNNs, but also on learning machines in which RNNs interact with external memory such as neural Turing machines, memory networks, and related memory architectures such as fast weight networks and neural stack machines.

This is the conference, and here's my talk (will do Google hangout, just as with my recent talks in Bern, Strasbourg, etc): Through a series of examples, we consider problems with classical hypothesis testing, whether performed using classical p-values or confidence intervals, Bayes factors, or Bayesian inference using noninformative priors. We locate the problem not in the use of any particular statistical method but rather with larger problems of deterministic thinking and a misguided version of Popperianism in which the rejection of a straw-man null hypothesis is taken as confirmation of a preferred alternative. We suggest solutions involving multilevel modeling and informative Bayesian inference. The post Hypothesis Testing is a Bad Idea (my talk at Warwick, England, 2:30pm Thurs 15 Sept) appeared first on Statistical Modeling, Causal Inference, and Social Science. The post Hypothesis Testing is a Bad Idea (my talk at Warwick, England, 2:30pm Thurs 15 Sept) appeared first on All About Statistics.