The tenth annual Microsoft eScience Workshop will be held in Beijing in conjunction with the IEEE International Conference on eScience. It will continue the ongoing dialogue centered on applications in broad areas of scientific investigation, such as environmental studies, bioinformatics, and climate understanding; present new results in data modeling; and provide an opportunity for open discourse on developments in urban computing.

Since its founding in 2008, Microsoft Research New England has pursued new, interdisciplinary areas of research that bring together traditional mathematical sciences, such as theoretical computer science, machine learning, physics, and mathematics itself, with other sciences, including social sciences such as empirical economics, sociology, and psychology, and the biological and biomedical sciences.

Programming today involves code editing mixed with bouts of debugging to get feedback on code execution. For programming to be more fluid, editing and debugging should occur not only at the same time, but also in the same space to quickly make use of the resulting live execution feedback. This paper describes how live feedback can be woven into the editor by making places in code execution, not just in code, navigable so evaluation results can be probed directly within the code editor. A pane aside the...

In this second workshop on Wearable Systems for Industrial Augmented Reality Applications we will discuss the following topics, related to wearable computing and AR technologies: core technologies, such as hardware, AR development kits or AR-enabled software; software architectures and applications concepts; as well as business ideas and case studies of AR systems within the industrial context.

End users have begun to incorporate cloud-based services into their collaborative practices. What spurs and constrains this adoption? Are the cloud services understood adequately and used effectively? How might we intervene to promote a better connection between user practices and cloud services? In this study, we focus on collaborative practices that surround the adoption, use, and understanding of two popular, but sometimes contrasting, cloud services for creating and sharing content: Dropbox and Google...

We propose a general database-driven framework for coherent synthesis of subject-specific scans of unavailable modality, which adopts and generalizes the patch-based label propagation (LP) strategy. While modality synthesis has received increased attention lately, current methods are mainly tailored to specific applications. On the other hand, the LP framework has been extremely successful for certain segmentation tasks, however, so far it has not been used for estimation of entities other than categorical...

Photoshopping of newlyweds, downloading the latest movies, teens flirting on social network sites and virtual gaming may seem like typical behavior in the West; yet in the context of a village in Mali or a slum in Mumbai, it is seen as unusual and perhaps an anomaly in their new media practice. In recent years, some studies (Ganesh, 2010; Mitra, 2005; Arora, 2010; 2012; Rangaswamy & Cutrell, 2012; Kavoori, Chadha & Arceneaux, 2006) have documented these leisure-oriented behaviors in the global south and...

We propose a method for multi-atlas label propagation based on encoding the individual atlases by randomized classification forests. Most current approaches perform a non-linear registration between all atlases and the target image, followed by a sophisticated fusion scheme. While these approaches can achieve high accuracy, in general they do so at high computational cost. This negatively affects the scalability to large databases and experimentation. To tackle this issue, we propose to use a small and...

Accurate localization and identification of vertebrae in spinal imaging is crucial for the clinical tasks of diagnosis, surgical planning, and post-operative assessment. The main difficulties for automatic methods arise from the frequent presence of abnormal spine curvature, small field of view, and image artifacts caused by surgical implants. Many previous methods rely on parametric models of appearance and shape whose performance can substantially degrade for pathological cases. We propose a robust...

Previous research has shown that low-literate users have difficulty using hierarchical information architectures and that a list design showing all items at once on a PC screen works best for search tasks. However, the limited screen space on phones makes it impossible to show more than a few items at once on a single screen. Does a hierarchical UI work better on a phone? In this study, we compared the performance of non-literate users from Bangalore, India, on a search task using a hierarchical UI (four...

A challenge in large vocabulary spoken language understanding (SLU) is robustness to automatic speech recognition (ASR) errors. The state of the art approaches for semantic parsing rely on using discriminative sequence classification methods, such as conditional random fields (CRFs). Most dialog systems employ a cascaded approach where the best hypotheses from the ASR system are fed into the following SLU system. In our previous work, we have proposed the use of lattices towards joint recognition and...

The expected download throughput in cellular networks (3G/LTE) is primarily a function of wireless link quality between the mobile device and the cell tower it is connected to. In addition, throughput also depends on the volume of background network traffic due to other active users in the same cell, commonly referred to as cellular load. As a result, even when a user experiences good signal condition, throughput can be low in a heavily loaded cell. While signal quality can be measured locally on a mobile...

This study focuses on teenage users of public internet access venues (PAVs) in low-income neighbourhoods of Cape Town. It documents their cultivation of detailed ICT repertoires to make the most of available ICTs. It highlights the continuing importance of PAVs as supplements for poorly equipped schools, and reveals the incompleteness of any supposed transition to mobile-only internet use. While the mobile internet is opening up opportunities for young people, its current form still conflicts with the easy...

We present a method for the analysis of functional properties of large-scale DNA strand displacement (DSD) circuits based on Satisfiability Modulo Theories that enables us to prove the functional correctness of DNA circuit designs for arbitrary inputs, and provides significantly improved scalability and expressivity over existing methods. We implement this method as an extension to the Visual DSD tool, and use it to formalize the behavior of a 4-bit square root circuit, together with the components used...

While video communication is becoming quite popular among remote friends and family, recent usage practices have been extending beyond just talking heads to remotely sharing an experience by doing an activity together. However, current video chat tools are aimed at sharing talking heads and need to be reconsidered to support remotely sharing activities. We explore a specific remote shared activity – watching video programs – through a three-phase study. We surveyed people’s interest in watching video...

Static Analysis Symposium

Near Field Communication (NFC) enables physically proximate devices to communicate over very over short ranges in a peer-to-peer manner, without incurring the overhead of any complex network configuration effort. However, the adoption of NFC-enabled applications has been stymied by the low levels of penetration of NFC hardware. In this paper, we address the challenge of enabling NFClike capability on the existing base of mobile phones. To this end, we develop Dhwani, a novel, acoustics-based NFC system...

Secure multi-party computation (MPC) allows a set of parties to compute a function of their inputs while preserving input privacy and correctness. MPC has been an active area of research of cryptography for over 30 years. The last decade has witnessed significant interest and advances in the applied aspects of MPC. This workshop will bring together researchers in security and cryptography to discuss recent advances, challenges and research directions related to applied secure computation.

Programming contest associated with the International Conference on Functional Programming (ICFP)

In this paper we show how to automatically induce non-linear features for machine translation. The new features are selected to approximately maximize a BLEU-related objective and decompose on the level of local phrases, which guarantees that the asymptotic complexity of machine translation decoding does not increase. We achieve this by applying gradient boosting machines (Friedman,2000) to learn new weak learners (features) in the form of regression trees, using a differentiable loss function related to...

In this paper we show how to automatically induce non-linear features for machine translation. The new features are selected to approximately maximize a BLEU-related objective and decompose on the level of local phrases, which guarantees that the asymptotic complexity of machine translation decoding does not increase. We achieve this by applying gradient boosting machines (Friedman,2000) to learn new weak learners (features) in the form of regression trees, using a differentiable loss function related to...

We introduce a social media text normalization system that can be deployed as a preprocessing step for Machine Translation and various NLP applications to handle social media text. The proposed system is based on unsupervised learning of the normalization equivalences from unlabeled text. The proposed approach uses Random Walks on a contextual similarity bipartite graph constructed from n-gram sequences on large unlabeled text corpus. We show that the proposed approach has a very high precision of (92.43)...

A novel deep architecture, the Tensor Deep Stacking Network (T-DSN), is presented. The T-DSN consists of multiple, stacked blocks, where each block contains a bilinear mapping from two hidden layers to the output layer, using a weight tensor to incorporate higher-order statistics of the hidden binary features. A learning algorithm for the T-DSN's weight matrices and tensors is developed and described, in which the main parameter estimation burden is shifted to a convex sub-problem with a closed-form...