Abstract Recognizing emotions from physiological signals has proven to be important in various scenarios. To assist in developing emotion recognizers, software frameworks and toolboxes have emerged, offering ready-to-use components. However,these have limitations regarding the type of physiological signals supported, the recognition steps covered, or the acquisition of multiple physiological signals. This paper presents metaFERA, an architectural meta-framework for creating software frameworks for end-to-end emotion recognition from physiological signals. The modularity and flexibility of the meta-framework and the resulting frameworks allow the fast prototyping of emotion recognition systems and experiments to test and validate new algorithms. To that end, metaFERA offers: (i) a set of pre-configured blocks to which we can add behavior to create framework components; (ii) an easy way to add behavior to the pre-configured blocks; (iii) a channel-based communication mechanism that transparently and efficiently supports the exchange of information between components; (iv) a simple and easy way to use and link components from a resulting framework to create applications. Additionally, we provide a set of Web services, already configured, to make the resulting recognition systems available as a service. To validate metaFERA, we created a framework for Electrodermal Activity, an emotion recognizer to identify high/low arousal using the aforementioned framework, and a layer to offer the recognizer as a service.

The success of supervised models for emotion recognition on images heavily depends on the availability of images properly annotated. Although millions of images are presently available, only a few are annotated with reliable emotional information. Current emotion recognition solutions either use large amounts of weakly-labeled web images, which often contain noise that is unrelated to the emotions of the image, or transfer learning, which usually results in performance losses. Thus, it would be desirable to know which images would be useful to be annotated to avoid an extensive annotation effort. In this paper, we propose a novel approach based on active learning to choose which images are more relevant to be annotated. Our approach dynamically combines multiple active learning strategies and learns the best ones (without prior knowledge of the best ones). Experiments using nine benchmark datasets revealed that: (i) active learning allows to reduce the annotation effort, while reaching or surpassing the performance of a supervised baseline with as little as 3% to 18% of the baseline's training set, in classification tasks; (ii) our online combination of multiple strategies converges to the performance of the best individual strategies, while avoiding the experimentation overhead needed to identify them.

Emotions have an important role in daily life, not only in human interaction, but also in decision-making processes, and in the perception of the world around us. Due to the recent interest shown by the research community in establishing emotional interactions between humans and computers, the identification of the emotional state of the former became a need. This can be achieved through multiple measures, such as subjective self-reports, autonomic and neurophysiological measurements. In the last years, Electroencephalography (EEG) received considerable attention from researchers, since it can provide a simple, cheap, portable, and ease-to-use solution for identifying emotions. In this paper, we present a survey of the neurophysiological research performed from 2009 to 2016, providing a comprehensive overview of the existing works in emotion recognition using EEG signals. We focus our analysis in the main aspects involved in the recognition process (e.g., subjects, features extracted, classifiers), and compare the works per them. From this analysis, we propose a set of good practice recommendations that researchers must follow to achieve reproducible, replicable, well-validated and high-quality results. We intend this survey to be useful for the research community working on emotion recognition through EEG signals, and in particular for those entering this field of research, since it offers a structured starting point.

In this paper, we introduce a novel gaze-only interaction technique called EyeLinks, which was designed i) to support various types of discrete clickables (e.g. textual links, buttons, images, tabs, etc.); ii) to be easy to learn and use; iii) to mitigate the inaccuracy of affordable eye trackers. Our technique uses a two-step fixation approach: first, we assign numeric identifiers to clickables in the region where users gaze at and second, users select the desired clickable by performing a fixation on the corresponding confirm button, displayed in a sidebar. This two-step selection enables users to freely explore Web pages, avoids the Midas touch problem and improves accuracy.

In this demo paper we describe PlaylistCreator, an assisted approach for supporting the creation of music playlists. Our solution allows creators to express song selection and browsing through a visual representation of their intents in a unified view, which relies on a set-based model for representing sources of songs. Creators can convey their purposes by seamlessly combining criteria for song selection from either manual or automatic sources, such as artists and albums, or similarity measures between artists or songs.

We describe a novel system for synthesising video choreography using sketched visual storyboards comprising human poses (stick men) and action labels. First, we describe an algorithm for searching archival dance footage using sketched pose. We match using an implicit representation of pose parsed from a mix of challenging low and high fidelity footage. In a training pre-process we learn a mapping between a set of exemplar sketches and corresponding pose representations parsed from the video, which are generalized at query-time to enable retrieval over previously unseen frames, and over additional unseen videos. Second, we describe how a storyboard of sketched poses, interspersed with labels indicating connecting actions, may be used to drive the synthesis of novel video choreography from the archival footage.

Music recommendation systems based on Collaborative Filtering methods have been extensively developed over the last years. Typically, they work by analyzing the past user-song relationships, and provide informed guesses based on the overall information collected from other users. Although the music listening behavior is a repetitive and time-dependent process, these methods have not taken this into account and only consider user-song interaction for recommendation. In this work, we explore the usage of temporal context and session diversity in Session-based Collaborative Filtering techniques for music recommendation. We compared two techniques to capture the users' listening patterns over time: one explicitly extracts temporal properties and session diversity, to group and compare the similarity of sessions, the other uses a generative topic modeling algorithm, which is able to implicitly model temporal patterns. We evaluated the developed algorithms by measuring the Hit Ratio, and the Mean Reciprocal Rank. Results reveal that the inclusion of temporal information, either explicitly or implicitly, increases significantly the accuracy of the recommendation, while compared to the traditional session-based CF.

Over the past years, music listening histories have become easily accessible due to the expansion of online lifelogging services. These histories represent the sequence of songs listen by users over time. Although this data contains intrinsic users' tastes and listening behaviors, it has been mainly used to personalize recommendations. Tools to help users exploring and reasoning about the information contained in the listening history, only recently have started to emerge. In this paper we describe a new visualization and exploration tool that allows users to interactively browse their listening histories, while leading them to identify listening trends and habits. Our solution combines a rich-featured timeline-based visualization, a set of synchronized-views and an interactive filtering mechanism to provide a flexible, effective and easy to use system for the analysis and knowledge exploration of listening histories. This was complemented with brushing and highlighting techniques to uncover listening trends about artists, albums and songs. Experimental evaluation with users revealed that they were able to complete all the requested tasks with a low error rate, and that they found the solution flexible and easy to use. Additionally, users were able to infer about their main life events and listening changes, which indicates that our combination of visualization techniques is effective in conveying relevant information about the listening habits.

Blogs are widely used today to publish information on a regular basis. However, users have difficulty in exploring their content and in discovering relevant information. This is due, among other things, to blogs rigid structure, with very long pages, and to the lack of mechanisms for effective navigation and exploration. To overcome these problems, we developed an exploration tool, to help users navigate, browse and visualize blogs. It was developed based on the following four design principles i) a blog should provide an overview of its activity and content to help users identify publication patterns and relevant entries; ii) blogs should present rich compact representations of entries to help readers anticipate the content of posts; iii) comments should have a relevant role, since they convey the social component of the blog; iv) blogs should offer an interactive filtering mechanism to help users explore and find relevant posts. A comparative evaluation with users confirmed the validity of the design principles, since our solution was more efficient, effective and usable than the usual blog interface and the Google Reader.

Although, color is one of the most visually distinguishable visual properties, color alone is not enough to describe the content of images. The spatial organization of the different color regions also play an important role. In this paper, we propose and evaluate a new descriptor that combines information about color and about its spatial arrangement in an image. Moreover, the mechanism used to compute the descriptor provides support for partial matching of images and for the development of efficient retrieval systems. We first describe the spatial arrangement of the color regions using a topological graph, where vertices represent the color regions and edges represent connections between regions and also the color differences between them. To compute the descriptor from this graph representation we use the spectral graph theory, avoiding the need for direct graph comparison. We performed various experimental evaluations to compare the accuracy of our new descriptor with descriptors based only on color, based only on topological information and a combination of both.

Nowadays, people spend time using services to track their music listening history. Although these services provide statistics and small graphics/charts, they are mainly used to record and to allow direct access to the information, not providing any visualization and exploration functionality. In this paper we describe a new approach for browsing and visualizing music listening histories, which combines a timeline-based visualization, with a set of synchronized-views and an interactive filtering mechanism to provide a flexible and easy to use solution. This was complemented with brushing and highlighting techniques that allow users to observe trends on artists, albums and tracks listening. Experimental evaluation with users revealed that they were able to complete all the proposed tasks with a low error rate, and that they found the solution easy to use. Moreover, users liked our approach for browsing and exploring listening histories, emphasizing its flexibility and effectiveness, and founding the full experience engaging and rewarding.

The contribution of this paper is a sketch parser able to recognize the several components of a skeleton described using the drawing of a stick-man. We describe the sketch parser in detail, and briefly outline how it is applied to form the front-end of a sketch based retrieval system capable of searching for human poses in archival dance footage.

In this paper we present an approach for interactive exploration of photo collections, which combines seamlessly multitouch gestures and sketches. Multitouch gestures are used to navigate through the photo collection, while sketches are used to filter the images presented to the user, according to their properties and metadata. The different gestures provided for interaction, their correspondent actions and how they are invoked are described in the remainder of this paper.

In this paper we present MuVis, an interactive visualization and exploration tool for large music collections, based on music content and metadata. We combined a user-centered design with three main components: information visualization techniques (based on semantic ordered treemaps), music information retrieval mechanisms (for semantic and content-based information extraction) and dynamic queries, to offer users a more efficient, flexible and yet, easy to use solution for browsing music collections and to create playlists. Preliminary results reveal that our solution is faster and easier to use than the Windows Media Player, allowing users to perform a more effective and fast navigation, while getting a deeper knowledge of their library. Satisfaction survey revealed that users liked our approach for browsing, filtering and creating playlists, while at the same time they were able to "re-discover" forgotten music, due to the similarity mechanisms incorporated in our solution.

With the growing popularity of digital cameras, the organization, browsing, management and grouping of photos become a problem of every photograph (professional or amateur), because their collections easily achieve the order of thousands. Here, we present a system to automate these processes, which relies on photo information, such as, semantic features (extracted from content), meta-information and low level.

These days there are a lot of vector drawings available for people to integrate into documents. These come in a variety of formats, such as Corel, Postscript, CGM, WMF and recently SVG. Typically, such ClipArt drawings tend to be archieved and accessed by categories (e.g. food, shapes, transportation, etc.). However, to find a drawing among hundreds of thousands is not an easy task. While text-driven attempts at classifying image data have been recently supplemented with query-by-image content, these have been developed for bitmap-type data and cannot handle vectorial information. In this paper we present an approach to allow indexing and retrieving vector drawings by content from large datasets. Our prototype can already handle databases with thousands of drawings using commodity hardware. Furthermore, preliminary usability assessments show promising results and suggest good acceptance of sketching as a query mechanism by users.

Many indexing approaches for high-dimensional data points have evolved into very complex and hard to code algorithms. Sometimes this complexity is not matched by increase in performance. Motivated by these ideas, we take a step back and look at simpler approaches to indexing multimedia data. In this paper we propose a simple, (not simplistic) yet efficient indexing structure for high-dimensional data Points of variable dimension, using dimension reduction. Our approach maps multidimensional points to a 1D line by computing their Euclidean Norm and use a B/sup +/-Tree to store data points. We exploit B/sup +/-Tree efficient sequential search to develop simple, yet performant methods to implement point, range and nearest-neighbor queries. To evaluate our technique we conducted a set of experiments, using both synthetic and real data. We analyze creation, insertion and query times as a function of data set size and dimension. Results so far show that our simple scheme outperforms current approaches, such as the Pyramid Technique, the A-Tree and the SR-Tree, for many data distributions. Moreover, our approach seems to scale better both with growing dimensionality and data set size, while exhibiting low insertion and search times.

Presents a simple method, based on fuzzy logic, to recognize multi-stroke sketches of geometric shapes and uni-stroke gestural commands. It uses temporal adjacency and global geometric properties of figures to recognize a simple vocabulary of geometric shapes drawn in different line styles. The geometric features used (convex hull, largest-area inscribed and smallest-area enclosing polygons, perimeter and area ratios) are invariant with rotation and scale of figures. Through experimental evaluation we have found the method very usable with acceptable recognition rates although the multi-stroke approach poses problems in choosing appropriate values for timeouts. Although we have privileged simplicity over robustness, the method has proved suitable for interactive applications.