Multi-Level Shape Representation Using Global Deformations and Locally Adaptive Finite Elements

We present a model-based method for the multi-level shape, pose estimation and abstraction of an object's surface from range data. The surface shape is estimated based on the parameters of a superquadric that is subjected to global deformations (tapering and bending) and a varying number of levels of local deformations. Local deformations are implemented using locally adaptive finite elements whose shape functions are piecewise cubic functions with C ¹ continuity. The surface pose is estimated based on the model's translational and rotational degrees of freedom. The algorithm first does a coarse fit, solving for a first approximation to the translation, rotation and global deformation parameters and then does several passes of mesh refinement, by locally subdividing triangles based on the distance between the given datapoints and the model. The adaptive finite element algorithm ensures that during subdivision the desirable finite element mesh generation properties of conformity, non-degeneracy and smoothness are maintained. Each pass of the algorithm uses physics-based modeling techniques to iteratively adjust the global and local parameters of the model in response to forces that are computed from approximation errors between the model and the data. We present results demonstrating the multi-level shape representation for both sparse and dense range data.     

An Adversarial Neuro-Tensorial Approach for Learning Disentangled Representations

International Journal of Computer Vision - Several factors contribute to the appearance of an object in a visual scene, including pose, illumination, and deformation, among others. Each factor...     

RED-Net: A Recurrent Encoder–Decoder Network for Video-Based Face Alignment

We propose a novel method for real-time face alignment in videos based on a recurrent encoder–decoder network model. Our proposed model predicts 2D facial point heat maps regularized by both detection and regression loss, while uniquely exploiting recurrent learning at both spatial and temporal dimensions. At the spatial level, we add a feedback loop connection between the combined output response map and the input, in order to enable iterative coarse-to-fine face alignment using a single network model, instead of relying on traditional cascaded model ensembles. At the temporal level, we first decouple the features in the bottleneck of the network into temporal-variant factors, such as pose and expression, and temporal-invariant factors, such as identity information. Temporal recurrent learning is then applied to the decoupled temporal-variant features. We show that such feature disentangling yields better generalization and significantly more accurate results at test time. We perform a comprehensive experimental analysis, showing the importance of each component of our proposed model, as well as superior results over the state of the art and several variations of our method in standard datasets.     

An adaptive memory programming framework for the resource-constrained project scheduling problem

The Resource-Constrained Project Scheduling Problem (RCPSP) is one of the most intractable combinatorial optimisation problems that combines a set of constraints and objectives met in a vast variety of applications and industries. Its solution raises major theoretical challenges due to its complexity, yet presenting numerous practical dimensions. Adaptive memory programming (AMP) is one of the most successful frameworks for solving hard combinatorial optimisation problems (e.g. vehicle routing and scheduling). Its success stems from the use of learning mechanisms that capture favourable solution elements found in high-quality solutions. This paper challenges the efficiency of AMP for solving the RCPSP, to our knowledge, for the first time in the literature. Computational experiments on well-known benchmark RCPSP instances show that the proposed AMP consistently produces high-quality solutions in reasonable computational times.     

Efficient Money Burning in General Domains

We study mechanism design where the objective is to maximize the residual surplus, i.e., the total value of the outcome minus the payments charged to the agents, by truthful mechanisms. The motivation comes from applications where the payments charged are not in the form of actual monetary transfers, but take the form of wasted resources. We consider a general mechanism design setting with m discrete outcomes and n multidimensional agents. We present two randomized truthful mechanisms that extract an O(logm) fraction of the maximum social surplus as residual surplus. The first mechanism achieves an O(logm)-approximation to the social surplus, which is improved to an O(1)-approximation by the second mechanism. An interesting feature of the second mechanism is that it optimizes over an appropriately restricted space of probability distributions, thus achieving an efficient tradeoff between social surplus and the total amount of payments charged to the agents.     

Supporting Creativity through the Interactive Exploratory Search Paradigm

In creative tasks, there is a need to explore the space of available information in order to come up with diverse views before converging to a solution. In such tasks, typical search engines that follow the direct search paradigm fail to inspire users. It is hypothesized that contrary to typical engines, interactive exploratory search, which aims at revealing latent, alternative directions in the information space enabling user orientation and engagement, is better suited to assist users in their quest for serendipitous discoveries and inspiration. In this study, an interactive exploratory search tool that combines diversification of content and sources with a user interface design that visualizes clues from the social chatter – generated with micro-blogging services such as Twitter – and lets users interactively explore the available information space is presented. A profiling service and recommendation module in charge of delivering personalized social content complements the setting. A pilot and two task-based user studies comparing our system to a query-based baseline indicate that our system significantly improves inspirational discoveries by providing access to more interesting and serendipitous information.     

Challenges and opportunities in renovating public sector information by enabling linked data and analytics

Linked Data has become the current W3C recommended approach for publishing data on the World Wide Web as it is sharable, extensible, and easily re-usable. An ecosystem of linked data hubs in the Public Sector has the potential to offer significant benefits to its consumers (other public offices and ministries, as well as researchers, citizens and SMEs), such as increased accessibility and re-use value of their data through the use of web-scale identifiers and easy interlinking with datasets of other public data providers. The power and flexibility of the schema-defying Linked Data, however, is counterbalanced by inborn factors that diminish the potential for cost-effective and efficient adoption by the Public Sector. The paper analyzes these challenges in view of the current state-of-the-art in linked data technologies and proposes a technical framework that aims to hide the underlying complexity of linked data while maintaining and promoting the interlinking capabilities enabled by the Linked Data Paradigm. The paper presents the innovations behind our proposed solutions as well as their advantages, especially for the non-expert users.     

Imputation of clinical covariates in time series

Machine Learning - Missing data is a common problem in longitudinal datasets which include multiple instances of the same individual observed at different points in time. We introduce a new...     

Vapor sorption in thin supported polymer films studied by white light interferometry

In the present study, we apply a white light interferometric methodology to study sorption of moisture and methanol vapor in thin films of poly(2-hydroxyethyl methacrylate) [PHEMA] and poly(methyl methacrylate) [PMMA], supported on oxidized silicon wafers. The measured equilibrium thickness expansion of each film, exposed to different activities of the vapor penetrant, is used to determine the sorption isotherm of the system. Results for relatively thick films (100 nm < L_o < 600 nm) are compared with corresponding literature data for bulk, free-standing films, obtained by direct gravimetric methods. Furthermore, PMMA films of thicknesses lower than 100 nm were employed in order to study the effect of the dry film's thickness, and of substrate, on fractional swelling.