In this work, we propose a novel automatic makeup detector and remover framework. For makeup detector, a locality-constrained low-rank dictionary learning algorithm is used to determine and locate the usage of cosmetics. For the challenging task of makeup removal, a locality-constrained coupled dictionary learning (LC-CDL) framework is proposed to synthesize non-makeup face, so that the makeup could be erased according to the style. Moreover, we build a stepwise makeup dataset (SMU) which to the best of our knowledge is the first dataset with procedures of makeup. This novel technology itself carries many practical applications, e.g. products recommendation for consumers; user-specified makeup tutorial; security applications on makeup face verification. Finally, our system is evaluated on three existing (VMU, MIW, YMU) and one own-collected makeup datasets. Experimental results have demonstrated the effectiveness of DL-based method on makeup detection. The proposed LC-CDL shows very promising performance on makeup removal regarding on the structure similarity. In addition, the comparison of face verification accuracy with presence or absence of makeup is presented, which illustrates an application of our automatic makeup remover system in the context of face verification with facial makeup.

In a wide range of statistical learning problems such as ranking, clustering or metric learning among others, the risk is accurately estimated by $U$-statistics of degree $d\geq 1$, i.e. functionals of the training data with low variance that take the form of averages over $k$-tuples. From a computational perspective, the calculation of such statistics is highly expensive even for a moderate sample size $n$, as it requires averaging $O(n^d)$ terms. This makes learning procedures relying on the optimization of such data functionals hardly feasible in practice. It is the major goal of this paper to show that, strikingly, such empirical risks can be replaced by drastically computationally simpler Monte-Carlo estimates based on $O(n)$ terms only, usually referred to as incomplete $U$-statistics, without damaging the $O_{\mathbb{P}}(1/\sqrt{n})$ learning rate of Empirical Risk Minimization (ERM) procedures. For this purpose, we establish uniform deviation results describing the error made when approximating a $U$-process by its incomplete version under appropriate complexity assumptions. Extensions to model selection, fast rate situations and various sampling techniques are also considered, as well as an application to stochastic gradient descent for ERM. Finally, numerical examples are displayed in order to provide strong empirical evidence that the approach we promote largely surpasses more naive subsampling techniques.

GettyFacebook CEO Mark Zuckerberg's 2016 New Years resolution is to create a virtual assistant for his home. Facebook is investing heavily in what many in the tech industry believe to be the next frontier of innovation, artificial intelligence. The largest social network on earth has a division of AI experts it calls FAIR. There's also a separate team called Applied Machine Learning, which focuses on "giving people communication superpowers through AI." Facebook clearly believes that AI is important to the company's future. Its employees are running 50x more AI experiments per day compared to last year.

Anticipating whether the car in front is going to take the next left or is slowing down is a fundamental part of driving, and key to not totalling your car. While keeping your eyes ahead should be second nature to those on the road, this most basic of tasks is still a challenge for driverless cars. But details have emerged of a patent filed by Google for its autonomous vehicles to detect and track brake and indicator lights of other cars on the road. Google has filed a patent for its autonomous vehicles to detect and track brake and indicator lights of other cars on the road. This will enable the driverless cars to better anticipate the movements of cars on the road. The technology would enable the driverless cars to anticipate the movements of cars on the road ahead using a forward-facing camera.

In theory, chatbots on Messenger would allow businesses to provide customer service without involving human workers. If you wanted to reach out to a business, you could do so via your Messenger app, rather than by looking up a phone number and calling. For example, Facebook showed off a chatbot for 1-800-Flowers.com that automatically takes orders via Messenger. It says things like: "White is a great choice! What is the recipient's name?"

For millions of people with swallowing disorders, preventing potentially deadly aspiration pneumonia requires following prescribed safe eating strategies. But adherence is poor, and caregivers’ ability to encourage adherence is limited by the onerous and socially aversive need to monitoring another’s eating. We have developed an early prototype for an intelligent assistant that monitors adherence and provides feedback to the patient, and tested monitoring precision with healthy subjects for one strategy called a “chin tuck.” Results indicate that adaptations of current generation machine vision and personal assistant technologies could effectively monitor chin tuck adherence, and suggest the feasibility of a more general assistant that encourages adherence to a wide range of safe eating strategies.

Constrained sampling and counting are two fundamental problems in artificial intelligence with a diverse range of applications, spanning probabilistic reasoning and planning to constrained-random verification. While the theory of these problems was thoroughly investigated in the 1980s, prior work either did not scale to industrial size instances or gave up correctness guarantees to achieve scalability. Recently, we proposed a novel approach that combines universal hashing and SAT solving and scales to formulas with hundreds of thousands of variables without giving up correctness guarantees. This paper provides an overview of the key ingredients of the approach and discusses challenges that need to be overcome to handle larger real-world instances.

The problem of designing recommender systems for scholarly article citations has been actively researched with more than 200 publications appearing in the last two decades. In spite of this, no definitive results are available about what approaches work best. Arguably the most important reason for this lack of consensus is the dearth of standardised test collections and evaluation protocols, such as those provided by TREC-like forums. CiteSeerX, a "scholarly big dataset" has recently become available. However, this collection provides only the raw material that is yet to be moulded into Cranfield style test collections. In this paper, we discuss the limitations of test collections used in earlier work, and describe how we used CiteSeerX to design a test collection with a well-defined evaluation protocol. The collection consists of over 600,000 research papers and over 2,500 queries. We report some preliminary experimental results using this collection, which are indicative of the performance of elementary content-based techniques. These experiments also made us aware of some shortcomings of CiteSeerX itself.

Retrieval systems for scholarly literature offer the ability for the scientific community to search, explore and download scholarly articles across various scientific disciplines. Mostly used by the experts in the particular field, these systems contain user community logs including information on user specific downloaded articles. In this paper we present a novel approach for automatically evaluating document similarity models in large collections of scholarly publications. Unlike typical evaluation settings that use test collections consisting of query documents and human annotated relevance judgments, we use download logs to automatically generate pseudo-relevant set of similar document pairs. More specifically we show that consecutively downloaded document pairs, extracted from a scholarly information retrieval (IR) system, could be utilized as a test collection for evaluating document similarity models. Another novel aspect of our approach lies in the method that we employ for evaluating the performance of the model by comparing the distribution of consecutively downloaded document pairs and random document pairs in log space. Across two families of similarity models, that represent documents in the term vector and topic spaces, we show that our evaluation approach achieves very high correlation with traditional performance metrics such as Mean Average Precision (MAP), while being more efficient to compute.

In this paper, we present an open-source emotion tracker and its innovative applications. Our tracker, EmoGram, tracks emotion changes for a sequence of textual units. It is versatile in terms of the textual unit (tweets, sentences in discourse, etc.) and also what constitutes the time sequence (timestamps of tweets, discourse nature of text, etc.). We demonstrate the utility of our system through our applications: a sequence of commentaries in cricket matches, a sequence of dialogues in a play, and a sequence of tweets related to the Maggi controversy in India in 2015. That one system can be used for these applications is the merit of EmoGram.