1,720,966 research outputs found
The Role of the Input in Natural Language Video Description
Full text linkNatural language video description (NLVD) has recently received strong interest in the computer vision, natural language processing (NLP), multimedia, and autonomous robotics communities. The state-of-the-art (SotA) approaches obtained remarkable results when tested on the benchmark datasets. However, those approaches poorly generalize to new datasets. In addition, none of the existing works focus on the processing of the input to the NLVD systems, which is both visual and textual. In this paper, an extensive study is presented to deal with the role of the visual input, evaluated with respect to the overall NLP performance. This is achieved by performing data augmentation of the visual component, applying common transformations to model camera distortions, noise, lighting, and camera positioning that are typical in real-world operative scenarios. A t-SNE-based analysis is proposed to evaluate the effects of the considered transformations on the overall visual data distribution. For this study, the English subset of the Microsoft Research Video Description (MSVD) dataset is considered, which is used commonly for NLVD. It was observed that this dataset contains a relevant amount of syntactic and semantic errors. These errors have been amended manually, and the new version of the dataset (called MSVD-v2) is used in the experimentation. The MSVD-v2 dataset is released to help to gain insight into the NLVD problem
A Model Reference Adaptive Control Approach for Uncertain Dynamical Systems with Strict Component-wise Performance Guarantees
No full textA critical problem for adaptive control systems is the characterization of the system re-sponse during transients. In fact a major issue in adaptive system design is the inability to achieve, a-priori, non-conservative user-defined performance guarantees. At present most of the available analysis tools provide performance bounds depending on the norm of uncertain quantities. Since it is extremely difficult to quantify these quantities, conservative upper bounds are used in their place; these, in turn, produce conservative performance bounds of limited practical utility. To face these problems some of the authors have recently introduced a set-theoretic adaptive controller based on generalized restricted potential functions. The key feature of this approach is that it allows the norm of the tracking error to be less than a-priori user-defined worst-case performance bound, and hence, it has the capability to enforce strict per-formance guarantees. Since this performance is expressed as function of the norm of the er-ror vector it is not possible to have the direct control on the amplitude of the single error components. In this paper the method is improved by allowing the control of the shape of the perfor-mance (ellipsoidal) set that is guaranteed to contain the tracking error trajectories. The de-sign problem is formalized as a linear optimization with LMI constraints that allows specify-ing independent componentwise requirements for the error components. Different linear optimization cost functions have been evaluated with the purpose of computing the largest ellipsoidal domain contained in an a-priori specified tracking error polyhedral domain and the smallest ellipsoidal domain containing an a-priori specified ellip-soidal initial condition set. A detailed simulation study in the aeronautic context has been used to highlight the efficacy of the method and the role of the different design parameters
A Comprehensive Case Study of Data-Driven Methods for Robust Aircraft Sensor Fault Isolation
No full textRecent catastrophic events in aviation have shown that current fault diagnosis schemes may not be enough to ensure a reliable and prompt sensor fault diagnosis. This paper describes a comparative analysis of consolidated data-driven sensor Fault Isolation (FI) and Fault Estimation (FE) techniques using flight data. Linear regression models, identified from data, are derived to build primary and transformed residuals. These residuals are then implemented to develop fault isolation schemes for 14 sensors of a semi-autonomous aircraft. Specifically, directional Mahalanobis distance-based and fault reconstruction-based techniques are compared in terms of their FI and FE performance. Then, a bank of Bayesian filters is proposed to compute, in flight, the fault belief for each sensor. Both the training and the validation of the schemes are performed using data from multiple flights. Artificial faults are injected into the fault-free sensor measurements to reproduce the occurrence of failures. A detailed evaluation of the techniques in terms of FI and FE performance is presented for failures on the air-data sensors, with special emphasis on the True Air Speed (TAS), Angle of Attack (AoA), and Angle of Sideslip (AoS) sensors
Experimental interval models for the robust Fault Detection of Aircraft Air Data Sensors
No full textIn this paper data-based approaches for a robust Fault Detection (FD) of the Air Data Sensors (ADS) including airspeed angles of attack and sideslip are proposed. Experimental Interval Models (IMs) have been considered for coping with modeling uncertainty and for providing interval estimations of the ADS signals. Specifically, a nonlinear-in-the-parameter Neural Network model has been introduced to characterize the nominal nonlinear response in the different phase of the flight, while model uncertainty is captured by an additional additive contribution provided by a linear in the parameters IM. The FD is achieved by verifying whether the measured ADS signals fall within time-varying bounds predicted by the nonlinear + IM. The IM identification has been formalized as a Linear Matrix Inequality (LMI) problem using as cost function the mean amplitude of the prediction interval and, as optimization variables, the amplitudes of the uncertain parameters of the model. The model identification was based on multi flight experimental data of a P92 Tecnam aircraft. The proposed method is compared with conventional FD schemes with fixed thresholds. Extensive validation tests have been conducted by injecting artificially additive hard and incipient failures on the ADS. The FD scheme has shown to be remarkably robust in all phases of the flight in terms of low false alarm rates while maintaining desirable detectability to faults. © 2018 Elsevier Lt
Visual Localization in the Presence of Appearance Changes Using the Partial Order Kernel
Full text linkVisual localization across seasons and under varying weather and lighting conditions is a challenging task in robotics. In this paper, we present a new sequence-based approach to visual localization using the Partial Order Kernel (POKer), a convolution kernel for string comparison, that is able to handle appearance changes and is robust to speed variations. We use multiple sequence alignment to construct directed acyclic graph representations of the database image sequences, where sequences of images of the same place acquired at different times are represented as alternative paths in a graph. We then use the POKer to compute the pairwise similarities between these graphs and the query image sequences obtained in a subsequent traversal of the environment, and match the corresponding locations. We evaluated our approach on a dataset which features extreme appearance variations due to seasonal changes. The results demonstrate the effectiveness of our approach, where it achieves
higher precision and recall than two state-of-the-art baseline method
Going Beyond Counting First Authors in Author Co-citation Analysis
Full text linkThe present study examines one of the fundamental aspects of author co-citation analysis (ACA) - the way co-citation
counts are defined. Co-citation counting provides the data on which all subsequent statistical analyses and mappings
are based, and we compare ACA results based on two different types of co-citation counting - the traditional type that
only counts the first one among a cited work's authors on the one hand and a non-traditional type that takes into
account the first 5 authors of a cited work on the other hand. Results indicate that the picture produced through this non-traditional author co-citation counting contains more coherent author groups and is therefore considerably clearer. However, this picture represents fewer specialties in the research field being studied than that produced through the traditional first-author co-citation counting when the same number of top-ranked authors is selected and analyzed. Reasons for these effects are discussed
Variations on the Author
Full text link“Variations on the Author” discusses two of Eduardo Coutinho’s recent films (Um Dia na Vida, from 2010, and Últimas Conversas, posthumously released in 2015) and their contribution to the general question of documentary authorship. The director’s filmography is characterized by a consistent yet self-effacing form of authorial self-inscription: Coutinho often features as an interviewer that rather than express opinions propels discourses; an interviewer that is good at listening. This mode of self-inscription characterizes him as an author who is not expressive but who is nonetheless markedly present on the screen. In Um Dia na Vida, however, Coutinho is completely absent form the image, while Últimas Conversas, on the contrary, includes a confessional prologue that moves the director from the margins to the center of his films. This article examines the ways in which these works stand out in the filmography of a director who offers new insights into the notion of cinematic authorship
- …
