Toronto, Aug 19 (IANS) Dodgy lip sync dubbing could soon become a thing of the past as researchers have developed an Artificial Intelligence (AI)-based system that can edit the facial expressions of actors to accurately match dubbed voices. The system, called Deep Video Portraits, can also be used to correct gaze and head pose in video conferencing, and enables new possibilities for video post-production and visual effects, according to the research presented at the SIGGRAPH 2018 conference in Vancouver, Canada. "This technique could also be used for post-production in the film industry where computer graphics editing of faces is already widely used in today's feature films," said study co-author Christian Richardt from the University of Bath in Britain. The researchers believe that the new system could help the film industry save time and reduce post-production costs. Unlike previous methods that are focused on movements of the face interior only, Deep Video Portraits can also animate the whole face including eyes, eyebrows, and head position in videos, using controls known from computer graphics face animation.

The graph represents a network of 2,124 Twitter users whose tweets in the requested range contained "#FinServ", or who were replied to or mentioned in those tweets. The network was obtained from the NodeXL Graph Server on Sunday, 19 August 2018 at 12:22 UTC. The requested start date was Sunday, 19 August 2018 at 00:01 UTC and the maximum number of days (going backward) was 14. The maximum number of tweets collected was 5,000. The tweets in the network were tweeted over the 7-day, 8-hour, 30-minute period from Saturday, 11 August 2018 at 15:30 UTC to Sunday, 19 August 2018 at 00:01 UTC.

On this episode of Recode Decode, hosted by Kara Swisher, Carnegie Mellon's Andrew Moore talks about the future of tech education as fields like artificial intelligence and machine learning take center stage. Moore, the dean of CMU's computer science school, says he's "concerned" that anti-immigrant fervor will deter the next generation of great computer scientists from coming to America, although CMU has not yet seen an impact on its application numbers. "I think it's short-term, and I haven't seen any craziness, though of course, I'm frightened that it'll happen -- on this question of getting really the strongest folks over," Moore said. "If we appear to have a society which doesn't welcome folks from elsewhere then of course any sane brilliant scientist will end up going to Canada or Singapore or Zurich because they'll be able to get the best of both worlds." "Once you're living in an academic community or in a software development office for an exciting company, usually in day-to-day interactions this doesn't come up," he added. "You're so focused on some particular mission. But that perception -- especially among someone who's maybe 16 or 17 in anywhere from Turkey to China to England -- is something I'm concerned about." On the new podcast, he also talks about the often-forgotten importance of electrical and computer engineers, who will develop the sensors that make machine learning advance; how educational programs have been complicit in the lack of diversity in tech; and why he's personally pessimistic that self-driving cars, one of Carnegie Mellon's areas of expertise, will be ready by the early 2020s, as some have predicted. You can listen to Recode Decode on Apple Podcasts, Spotify, Pocket Casts, Overcast or wherever you listen to podcasts. Below, we've shared a lightly edited transcript of Kara's full conversation with Andrew. Kara Swisher: Today, I'm delighted to have Andrew Moore on the podcast. He's the dean of Carnegie Mellon's School of Computer Science, which was ranked No. 1 in the world by U.S. News and World Report. And he was previously a vice president of engineering at Google where he was in charge of Google Shopping. Andrew Moore: Happy to be here, thank you. I wanna get your background. I've had various computer scientists on the show who are teaching and like that, and I'd love to get sort of the academic perspective, but you've been in the fray, also. So just let's give your background, where you came from and how you got to Carnegie Mellon and then we'll talk about what's going on there. I grew up in a seaside town called Bournemouth in South of England, and there, in the late '80s, I really got into creating video games, like a lot of kids at the time.

Deep Neural Networks are powerful models that attained remarkable results on a variety of tasks. These models are shown to be extremely efficient when training and test data are drawn from the same distribution. However, it is not clear how a network will act when it is fed with an out-of-distribution example. In this work, we consider the problem of out-of-distribution detection in neural networks. We propose to use multiple semantic dense representations instead of sparse representation as the target label. Specifically, we propose to use several word representations obtained from different corpora or architectures as target labels. We evaluated the proposed model on computer vision, and speech commands detection tasks and compared it to previous methods. Results suggest that our method compares favorably with previous work. Besides, we present the efficiency of our approach for detecting wrongly classified and adversarial examples.

A large number of papers have introduced novel machine learning and feature extraction methods for automatic classification of AD. However, they are difficult to reproduce because key components of the validation are often not readily available. These components include selected participants and input data, image preprocessing and cross-validation procedures. The performance of the different approaches is also difficult to compare objectively. In particular, it is often difficult to assess which part of the method provides a real improvement, if any. We propose a framework for reproducible and objective classification experiments in AD using three publicly available datasets (ADNI, AIBL and OASIS). The framework comprises: i) automatic conversion of the three datasets into BIDS format, ii) a modular set of preprocessing pipelines, feature extraction and classification methods, together with an evaluation framework, that provide a baseline for benchmarking the different components. We demonstrate the use of the framework for a large-scale evaluation on 1960 participants using T1 MRI and FDG PET data. In this evaluation, we assess the influence of different modalities, preprocessing, feature types, classifiers, training set sizes and datasets. Performances were in line with the state-of-the-art. FDG PET outperformed T1 MRI for all classification tasks. No difference in performance was found for the use of different atlases, image smoothing, partial volume correction of FDG PET images, or feature type. Linear SVM and L2-logistic regression resulted in similar performance and both outperformed random forests. The classification performance increased along with the number of subjects used for training. Classifiers trained on ADNI generalized well to AIBL and OASIS. All the code of the framework and the experiments is publicly available at: https://gitlab.icm-institute.org/aramislab/AD-ML.

This paper describes ASAL a new active learning strategy that uses uncertainty sampling, adversarial sample generation and sample matching. Compared to traditional pool-based uncertainty sampling strategies, ASAL synthesizes uncertain samples instead of performing an exhaustive search in each active learning cycle. Then, the sample matching efficiently selects similar samples from the pool. We present a comprehensive set of experiments on MNIST and CIFAR-10 and show that ASAL outperforms similar methods and clearly exceeds passive learning. To the best of our knowledge this is the first pool-based adversarial active learning technique and the first that is applied for multi-label classification using deep convolutional classifiers.

The goal of a recommender system is to show its users items that they will like. In forming its prediction, the recommender system tries to answer: "what would the rating be if we 'forced' the user to watch the movie?" This is a question about an intervention in the world, a causal question, and so traditional recommender systems are doing causal inference from observational data. This paper develops a causal inference approach to recommendation. Traditional recommenders are likely biased by unobserved confounders, variables that affect both the "treatment assignments" (which movies the users watch) and the "outcomes" (how they rate them). We develop the deconfounded recommender, a strategy to leverage classical recommendation models for causal predictions. The deconfounded recommender uses Poisson factorization on which movies users watched to infer latent confounders in the data; it then augments common recommendation models to correct for potential confounding bias. The deconfounded recommender improves recommendation and it enjoys stable performance against interventions on test sets.

We study the learning capacity of empirical risk minimization with regard to the squared loss and a convex hypothesis class consisting of linear functions. While these types of estimators were originally designed for noisy linear regression problems, it recently turned out that they are in fact capable of handling considerably more complicated situations, involving highly non-linear distortions. This work intends to provide a comprehensive explanation of this somewhat astonishing phenomenon. At the heart of our analysis stands the mismatch principle, which is a simple, yet generic recipe to establish theoretical error bounds for empirical risk minimization. The scope of our results is fairly general, permitting arbitrary sub-Gaussian input-output pairs, possibly with strongly correlated feature variables. Noteworthy, the mismatch principle also generalizes to a certain extent the classical orthogonality principle for ordinary least squares. This adaption allows us to investigate problem setups of recent interest, most importantly, high-dimensional parameter regimes and non-linear observation processes. In particular, our theoretical framework is applied to various scenarios of practical relevance, such as single-index models, variable selection, and strongly correlated designs. We thereby demonstrate the key purpose of the mismatch principle, that is, learning (semi-)parametric output rules under large model uncertainties and misspecifications.

In this dissertation, I derive a new method to estimate the Vapnik-Chervonenkis Dimension (VCD) for the class of linear functions. This method is inspired by the technique developed by Vapnik et al. Vapnik et al. (1994). My contribution rests on the approximation of the expected maximum difference between two empirical Losses (EMDBTEL). In fact, I use a cross-validated form of the error to compute the EMDBTEL, and I make the bound on the EMDBTEL tighter by minimizing a constant in of its right upper bound. I also derive two bounds for the true unknown risk using the additive (ERM1) and the multiplicative (ERM2) Chernoff bounds. These bounds depend on the estimated VCD and the empirical risk. These bounds can be used to perform model selection and to declare with high probability, the chosen model will perform better without making strong assumptions about the data generating process (DG). I measure the accuracy of my technique on simulated datasets and also on three real datasets. The model selection provided by VCD was always as good as if not better than the other methods under reasonable conditions.

Artificial Intelligence in healthcare is a new and exciting frontier and the possibilities are endless. With deep learning approaches beating human performances in many areas, the logical next step is to attempt their application in the health space. For these and other Machine Learning approaches to produce good results and have their potential realized, the need for, and importance of, large amounts of accurate data is second to none. This is a challenge faced by many industries and more so in the healthcare space. We present an approach of using Variational Autoencoders (VAE's) as an approach to generating more data for training deeper networks, as well as uncovering underlying patterns in diagnoses and the patients suffering from them. By training a VAE, on available data, it was able to learn the latent distribution of the patient features given the diagnosis. It is then possible, after training, to sample from the learnt latent distribution to generate new accurate patient records given the patient diagnosis.