Assessing uncertainty is an important step towards ensuring the safety and reliability of machine learning systems. Existing uncertainty estimation techniques may fail when their modeling assumptions are not met, e.g. when the data distribution differs from the one seen at training time. Here, we propose techniques that assess a classification algorithm’s uncertainty via calibrated probabilities (i.e. probabilities that match empirical outcome frequencies in the long run) and which are guaranteed to be reliable (i.e. accurate and calibrated) on out-of-distribution input, including input generated by an adversary. This represents an extension of classical online learning that handles uncertainty in addition to guaranteeing accuracy under adversarial assumptions. We establish formal guarantees for our methods, and we validate them on two real-world problems: question answering and medical diagnosis from genomic data.

The exponential growth of information on Community-based Question Answering (CQA) sites has raised the challenges for the accurate matching of high-quality answers to the given questions. Many existing approaches learn the matching model mainly based on the semantic similarity between questions and answers, which can not effectively handle the ambiguity problem of questions and the sparsity problem of CQA data. In this paper, we propose to solve these two problems by exploiting users' social contexts. Specifically, we propose a novel framework for CQA task by exploiting both the question-answer content in CQA site and users' social contexts. The experiment on real-world dataset shows the effectiveness of our method.

We propose a scalable approach to learn video-based question answering (QA): to answer a free-form natural language question about the contents of a video. Our approach automatically harvests a large number of videos and descriptions freely available online. Then, a large number of candidate QA pairs are automatically generated from descriptions rather than manually annotated. Next, we use these candidate QA pairs to train a number of video-based QA methods extended from MN (Sukhbaatar et al. 2015), VQA (Antol et al. 2015), SA (Yao et al. 2015), and SS (Venugopalan et al. 2015). In order to handle non-perfect candidate QA pairs, we propose a self-paced learning procedure to iteratively identify them and mitigate their effects in training. Finally, we evaluate performance on manually generated video-based QA pairs. The results show that our self-paced learning procedure is effective, and the extended SS model outperforms various baselines.

Nowadays the community-based question answering (CQA) sites become the popular Internet-based web service, which have accumulated millions of questions and their posted answers over time. Thus, question answering becomes an essential problem in CQA sites, which ranks the high-quality answers to the given question. Currently, most of the existing works study the problem of question answering based on the deep semantic matching model to rank the answers based on their semantic relevance, while ignoring the authority of answerers to the given question. In this paper, we consider the problem of community-based question answering from the viewpoint of asymmetric multi-faceted ranking network embedding. We propose a novel asymmetric multi-faceted ranking network learning framework for community-based question answering by jointly exploiting the deep semantic relevance between question-answer pairs and the answerers' authority to the given question. We then develop an asymmetric ranking network learning method with deep recurrent neural networks by integrating both answers' relative quality rank to the given question and the answerers' following relations in CQA sites. The extensive experiments on a large-scale dataset from a real world CQA site show that our method achieves better performance than other state-of-the-art solutions to the problem.

Artificial intelligence has been a dream in technology ever since Alan Turing first wrote his seminal paper, Computing Machinery and Intelligence, Now, thanks to advances in hardware power and algorithm design, AI is a growth industry – and it has no shortage of vocal advocates. These are some of the most vocal and influential leaders working on artificial intelligence, robotics, chat bots, virtual reality, the ethics of autonomous software and vehicles and more. Organizes the London.AI meet up and the annual Playfair AI Summit. Thanks to all who kicked off discussion on the back of my piece on "6 areas of #AI/ML to watch closely" Keep going! Designs intelligent systems into working AI systems to help understand natural intelligence.

If it wasn't already clear that AI-powered voice assistants are ready for the workplace, it is now. IBM is not only launching Watson for Cybersecurity, a cognitive computing service that parses legions of security reports to extract relevant info, but is unveiling an experimental voice helper to go along with it. Havyn lets digital defense experts ask for threat updates and recommended solutions when it would otherwise be too time-consuming. If security analysts are already hip-deep in work, they don't have to sidetrack themselves with a new research path when Havyn can produce a useful answer in seconds. The combo could be particularly helpful given Watson's depth.

PubNub BLOCKS enables you to process your data mid-stream, to execute functions on your data in motion. This is huge, because you no longer need to spin up and manage new servers to run a simple function. It's all done in the network. That's why we say that PubNub is a programmable network. IBM Watson is a powerful technology that brings cognition to applications, with the ability to understand, reason, learn, and interact. I like to say it gives your application a brain, extending the capabilities of your application beyond simply interacting through a set of rules.

Several deep learning models have been proposed for question answering. However, due to their single-pass nature, they have no way to recover from local maxima corresponding to incorrect answers. To address this problem, we introduce the Dynamic Coattention Network (DCN) for question answering. The DCN first fuses co-dependent representations of the question and the document in order to focus on relevant parts of both. Then a dynamic pointing decoder iterates over potential answer spans. This iterative procedure enables the model to recover from initial local maxima corresponding to incorrect answers. On the Stanford question answering dataset, a single DCN model improves the previous state of the art from 71.0% F1 to 75.9%, while a DCN ensemble obtains 80.4% F1.

The tax code is more than 74,000 pages long, and there are thousands of new changes made each year that impact a client's tax outcome. H&R Block with Watson combines the expertise of 70,000 tax pros with the powerful technology of Watson to help ensure clients get back what they deserve. It's a first in the tax prep industry, and it represents H&R Block's most personalized tax experience ever. To get there, H&R Block tax professionals and IBM development teams are training Watson on the language of taxes. They're first applying the technology to the thousands of tax-related questions and topics discussed with an H&R Block Tax Pro during the return filing process.

From visual recognition to speech-to- text, the technology landscape continues to transform itself and it's happening rapidly. In 2017, the adoption and application of artificial intelligence is a more than just a far reaching dream, but a reality for most technology users. Not only is it critical that we identity these trends, but also build a workforce that adapt to these changes and build the new technologies that will advance society. We kicked off in New York City, hosting more than 40 coders, developers, early adopters, and those just interested in cognitive technology. All attendees were encouraged to take a coding course on the Learning Lab to prepare them for the event.