Catálogo de publicaciones - libros

This paper summarizes an invited talk given at the 9th International Conference on Intelligent Human Computer Interaction (December 2017, Paris). Algorithms have revolutionized almost every field of manufacturing and engineering. Is the design of user interfaces the next? This talk will give an overview of what future holds for algorithmic methods in this space. I introduce the idea of using predictive models and simulations of end-user behavior in combinatorial optimization of user interfaces, as well as the contributions that inverse modeling and interactive design tools make. Several research results are presented from gesture design to keyboards and web pages. Going beyond combinatorial optimization, I discuss self-optimizing or “autonomous” UI design agents.

Using a dot-matrix display, it is possible to present geometrical shapes with different rendering methods: solid shapes, empty shapes, vibrating shapes, etc. An open question is then: This paper presents results of a user study that we have conducted to address this question. Using a 60 * 60 dot-matrix display, we asked 40 participants to recognize 6 different geometrical shapes (square, circle, simple triangle, right triangle, diamond and cross) within the shortest possible time. Six different methods to render the shapes were tested depending on the rendering of shape’s outline and inside: static outline combined with static or vibrant or empty inside, and vibrating outline combined with static or vibrant or empty inside. The results show that squares, right triangles, and crosses are more quickly recognized than circles, diamonds, and simple triangles. Furthermore, the best rendering method is the one that combines static outline with empty inside.

We present a pipeline for recognizing dynamic freehand gestures on mobile devices based on extracting depth information coming from a single Time-of-Flight sensor. Hand gestures are recorded with a mobile 3D sensor, transformed frame by frame into an appropriate 3D descriptor and fed into a deep LSTM network for recognition purposes. LSTM being a recurrent neural model, it is uniquely suited for classifying explicitly time-dependent data such as hand gestures. For training and testing purposes, we create a small database of four hand gesture classes, each comprising 40  150 3D frames. We conduct experiments concerning execution speed on a mobile device, generalization capability as a function of network topology, and classification ability ‘ahead of time’, i.e., when the gesture is not yet completed. Recognition rates are high (>95%) and maintainable in real-time as a single classification step requires less than 1 ms computation time, introducing freehand gestures for mobile systems.

In recent years, unmanned aerial vehicles have received a significant attention in the research community, due to their adaptability in different applications, such as surveillance, disaster response, traffic monitoring, transportation of goods, first aid, etc. Nowadays, even though UAVs can be equipped with some autonomous capabilities, they often operate in high uncertainty environments in which supervisory systems including human in the control loop are still required. Systems envisaging decision-making capabilities and equipped with flexible levels of autonomy are needed to support UAVs controllers in monitoring operations. The aim of this paper is to build an adjustable autonomy system able to assist UAVs controllers by predicting mental workload changes when the number of UAVs to be monitored highly increases. The proposed system adjusts its level of autonomy by discriminating situations in which operators’ abilities are sufficient to perform UAV supervision tasks from situations in which system suggestions or interventions may be required. Then, a user study was performed to create a mental-workload prediction model based on operators’ cognitive demand in drone monitoring operations. The model is exploited to train the system developed to infer the appropriate level of autonomy accordingly. The study provided precious indications to be possibly exploited for guiding next developments of the adjustable autonomy system proposed.

The advancement in brain-computer interface systems (BCIs) gives a new hope to people with special needs in restoring their independence. Since, BCIs using motor imagery (MI) rhythms provides high degree of freedom, it is been used for many real-time applications, especially for locked-in people. The available BCIs using MI-based EEG signals usually makes use of spatial filtering and powerful classification methods to attain better accuracy and performance. Inter-subject variability and speed of the classifier is still a issue in MI-based BCIs. To address the aforementioned issues, in this work, we propose a new classification method, spatial filtering based sparsity (SFS) approach for MI-based BCIs. The proposed method makes use of common spatial pattern (CSP) to spatially filter the MI signals. Then frequency bandpower and wavelet features from the spatially filtered signals are used to bulid two different over-complete dictionary matrix. This dictionary matrix helps to overcome the issue of inter-subject variability. Later, sparse representation based classification is carried out to classify the two-class MI signals. We analysed the performance of the proposed approach using publicly available MI dataset IVa from BCI competition III. The proposed SFS method provides better classification accuracy and runtime than the well-known support vector machine (SVM) and logistic regression (LR) classification methods. This SFS method can be further used to develop a real-time application for people with special needs.

The increasing popularity of unmanned aerial vehicles (UAVs) in critical applications makes supervisory systems based on the presence of human in the control loop of crucial importance. In UAV-traffic monitoring scenarios, where human operators are responsible for managing drones, systems flexibly supporting different levels of autonomy are needed to assist them when critical conditions occur. The assessment of UAV controllers’ performance thus their mental workload may be used to discriminate the level and type of automation required. The aim of this paper is to build a mental-workload prediction model based on UAV operators’ cognitive demand to support the design of an adjustable autonomy supervisory system. A classification and validation procedure was performed to both categorize the cognitive workload measured by ElectroEncephaloGram signals and evaluate the obtained patterns from the point of view of accuracy. Then, a user study was carried out to identify critical workload conditions by evaluating operators’ performance in accomplishing the assigned tasks. Results obtained in this study provided precious indications for guiding next developments in the field.

In the classification of multichannel electroencephalograph (EEG) based BCI studies, the spatial and spectral information related to brain activities associated with BCI paradigms are usually pre-determined as default without speculation, which can lead to loses effects in practical applications due to individual variability across different subjects. Recent studies have shown that feature combination of each specifically tailored for different physiological phenomena such as Readiness Potential (RP) and Event Related Desynchronization (ERD) might benefit BCI making it robust against artifacts. Hence, the objective is to design a CSSBP with combined feature vectors, where the signal is divided into several sub bands using a band pass filter, and this channel and frequency configurations are then modeled as preconditions before learning base learners and introducing a new heuristic of stochastic gradient boost for training the base learners under these preconditions. Results showed that Boosting approach using feature combination clearly outperformed the state-of-the-art algorithms, and improved the classification performance, resulting in increased robustness.

Displaying and interacting with Cascaded Menus on mobile phones is challenging due to the limited screen real estate. In this paper, we propose the – U-shaped layout displayed along the edges of the screen. Through the use of transparency and minimum screen space, the Edge Menu can be overlaid on top of existing items on the screen. We evaluated the suitability of two versions of the Edge Menu: List and Nested Menus. We compared the performance of the Edge Menu to the traditional Linear Menu. We conducted three studies and revealed that Edge Menu can support the use of single hand and both hands, it outperforms the regular Linear Menu, and is in average 38.5% faster for Single hand usage, and 40% faster for Dual hands usage. Edge Menu using both hands is in average 7.4% faster than Edge Menu using Single hand. Finally, the Edge Menu in Nested Menus shown to be faster than Linear Menus in Nested Menus with 22%–36%.

Video lectures form a primary part of MOOC instruction delivery design. They serve as gateways to draw students into the course. In going over these videos accumulating knowledge, there is a high occurrence of cases [] where the learner forgets about some of the concepts taught and focus more on what is the minimum amount knowledge needed to carry forward to attempt the quizzes and pass. This is a step backward when we are concerned with giving the learner a learning outcome that seems to bridge the gap between what he knew before and after the course to completion. To address this issue, we are proposing an interaction model that enables the learner to promptly take notes as and when the video is being viewed. The work contains a functional prototype of the application for taking personalized notes from MOOC contents. The work [] is an integration of several world leading MOOC providers content using application program interface(API) and a customize interface module for searching courses from multiple MOOC providers as COURSEEKA and personalised note taking module as MOOKbook. This paper largely focuses on a learner’s perspective towards video based lectures and interaction to find the enhancements in interaction with longer retention of MOOC contents.

In education, some students lack language comprehension, language production and language acquisition skills. In this paper we extracted several psycholinguistics features broadly grouped into lexical and morphological complexity, syntactic complexity, production units, syntactic pattern density, referential cohesion, connectives, amounts of coordination, amounts of subordination, LSA, word information, and readability from students’ summary speech transcripts. Using these Coh-Metrix features, comprehenders are classified into two groups: poor comprehender and proficient comprehender. It is concluded that a computational model can be implemented using a reduced set of features and the results can be used to help poor reading comprehenders for improving their cognitive reading skills.