Abstract
Bidirectional transformations, in particular lenses, are programs with a forward get transformation and a backward putback transformation that keep source and view data types synchronized. Several bidirectional programming languages exist to aid programmers in writing a (sort of) forward transformation, and deriving a backward transformation for free. However, the maintainability offered by such languages comes at the cost of expressiveness and (more importantly) predictability because the ambiguity of synchronization -handled by the putback transformation- is solved by default strategies over which programmers have little control. In this paper, we argue that controlling such ambiguity is essential for bidirectional transformations and propose a novel language in which programmers write a (sort of) putback transformation, and get the unique get transformation for free. Like traditional bidirectional languages, our put-oriented language allows reasoning about the correctness of defined transformations from the properties of their building blocks. But it allows programmers to describe the behavior of a bidirectional transformation much more precisely, while retaining the maintainability of writing a single program. We demonstrate the practical power of the new approach through a series of examples, ranging from simple ones that illustrate traditional lenses to complex ones for which our putback-based approach is central to specifying nontrivial update strategies. Categories and Subject Descriptors D.1.1 [Programming Techniques]: Applicative (Functional) Programming; D.3.1 [Programming Languages]: Formal Definitions and Theory; F.3.2 [Logics and Meanings of Programs]: Semantics of Programming Languages-Algebraic approaches to semantics.

Abstract
Different XML formats are widely used for data exchange and processing, being often necessary to mutually convert between them. Standard XML transformation languages, like XSLT or XQuery, are unsatisfactory for this purpose since they require writing a separate transformation for each direction. Existing bidirectional transformation languages mean to cover this gap, by allowing programmers to write a single program that denotes both transformations. However, they often 1) induce a more cumbersome programming style than their traditionally unidirectional relatives, to establish the link between source and target formats, and 2) offer limited configurability, by making implicit assumptions about how modifications to both formats should be translated that may not be easy to predict. This paper proposes a bidirectional XML update language called BIFLUX (BIdirectional FunctionaL Updates for XML), inspired by the FLUX XML update language. Our language adopts a novel bidirectional programming by update paradigm, where a program succinctly and precisely describes how to update a source document with a target document, in an intuitive way, such that there is a unique "inverse" source query for each update program. BIFLUX extends FLUX with bidirectional actions that describe the connection between source and target formats. We introduce a core BIFLUX language, with a clear and well-behaved bidirectional semantics and a decidable static type system based on regular expression types.

Abstract
Identifying bad design patterns in software is a successful and inspiring research trend. While these patterns do not necessarily correspond to software errors, the fact is that they raise potential problematic issues, often referred to as code smells, and that can for example compromise maintainability or evolution. The identification of code smells in spreadsheets, which can be viewed as software development environments for non-professional programmers, has already been the subject of confluent researches by different groups. While these research groups have focused on detecting smells on concrete spreadsheets, or spreadsheet instances, in this paper we propose a comprehensive set of smells for abstract representations of spreadsheets, or spreadsheet models. We also propose a set of refactorings suggesting how spreadsheet models can become simpler to understand, manipulate and evolve. Finally we present the integration of both smells and refactorings under the MDSheet framework.

Abstract
Traffic sampling is seen as a mandatory solution to cope with the huge amount of traffic traversing network devices. Despite the substantial research work in the area, improving the versatility of adjusting sampling to the wide variety of foreseeable measurement scenarios has not been targeted so far. This motivates the development of an encompassing measurement model based on traffic sampling able to support a large range of network management activities, in a scalable way. The design of this model involves identifying sampling techniques through its components rather than a closed unit, allowing to address issues such as flexibility, estimation accuracy, data overhead and computational weight within a narrower and simpler scope. This paper concretises these ideas presenting a modular and self-configurable measurement architecture based on sampling, a framework implementing sampling inherent pieces, and provides first results when deploying the proposed concepts in real traffic scenarios.

Abstract
Within network measurement context, traffic sampling has been targeted as a promising solution to cope with the huge amount of traffic traversing network devices as only a subset of packets is elected for analysis. Although this brings an evident advantage to measurement overhead, the computational burden of performing sampling tasks in network equipment may overshadow the potential benefits of sampling. Attending that sampling techniques evince distinct temporal and spatial characteristics in handling traffic, this paper is focused on studying the computational weight of current and emerging techniques in terms of memory consumption, CPU load and data volume. Furthermore, the accuracy of these techniques in estimating network parameters such as throughput is evaluated. A sampling framework has also been implemented in order to provide a versatile and fair platform for carrying out the testing and comparison process.

Abstract
In this paper we present a robust and lightweight method for the automatic fitting of deformable 3D face models on facial images. Popular fitting techniques such as those based on statistical models of shape and appearance require a training stage based on a set of facial images and their corresponding facial landmarks, which have to be manually labeled. Therefore, new images in which to fit the model cannot differ too much in shape and appearance (including illumination variation, facial hair, wrinkles, etc.) from those used for training. By contrast, our approach can fit a generic face model in two steps: (1) the detection of facial features based on local image gradient analysis and (2) the backprojection of a deformable 3D face model through the optimization of its deformation parameters. The proposed approach can retain the advantages of both learning-free and learning-based approaches. Thus, we can estimate the position, orientation, shape and actions of faces, and initialize user-specific face tracking approaches, such as Online Appearance Models (OAMs), which have shown to be more robust than generic user tracking approaches. Experimental results show that our method outperforms other fitting alternatives under challenging illumination conditions and with a computational cost that allows its implementation in devices with low hardware specifications, such as smartphones and tablets. Our proposed approach lends itself nicely to many frameworks addressing semantic inference in face images and videos.