Probabilistic consensus clustering using evidence accumulation

Machine Learning - Clustering ensemble methods produce a consensus partition of a set of data points by combining the results of a collection of base clustering algorithms. In the evidence...     

Assessment of crowdsourcing and gamification loss in user-assisted object segmentation

There has been a growing interest in applying human computation – particularly crowdsourcing techniques – to assist in the solution of multimedia, image processing, and computer vision problems which are still too difficult to solve using fully automatic algorithms, and yet relatively easy for humans. In this paper we focus on a specific problem – object segmentation within color images – and compare different solutions which combine color image segmentation algorithms with human efforts, either in the form of an explicit interactive segmentation task or through an implicit collection of valuable human traces with a game. We use Click’n’Cut, a friendly, web-based, interactive segmentation tool that allows segmentation tasks to be assigned to many users, and Ask’nSeek, a game with a purpose designed for object detection and segmentation. The two main contributions of this paper are: (i) We use the results of Click’n’Cut campaigns with different groups of users to examine and quantify the crowdsourcing loss incurred when an interactive segmentation task is assigned to paid crowd-workers, comparing their results to the ones obtained when computer vision experts are asked to perform the same tasks. (ii) Since interactive segmentation tasks are inherently tedious and prone to fatigue, we compare the quality of the results obtained with Click’n’Cut with the ones obtained using a (fun, interactive, and potentially less tedious) game designed for the same purpose. We call this contribution the assessment of the gamification loss, since it refers to how much quality of segmentation results may be lost when we switch to a game-based approach to the same task. We demonstrate that the crowdsourcing loss is significant when using all the data points from workers, but decreases substantially (and becomes comparable to the quality of expert users performing similar tasks) after performing a modest amount of data analysis and filtering out of users whose data are clearly not useful. We also show that – on the other hand – the gamification loss is significantly more severe: the quality of the results drops roughly by half when switching from a focused (yet tedious) task to a more fun and relaxed game environment.     

Landmark-based music recognition system optimisation using genetic algorithms

Audio fingerprinting allows us to label an unidentified music fragment within a previously generated database. The use of spectral landmarks aims to obtain a robustness that lets a certain level of noise be present in the audio query. This group of audio identification algorithms holds several configuration parameters whose values are usually chosen based upon the researcher’s knowledge, previous published experimentation or just trial and error methods. In this paper we describe the whole optimisation process of a Landmark-based Music Recognition System using genetic algorithms. We define the actual structure of the algorithm as a chromosome by transforming its high relevant parameters into various genes and building up an appropriate fitness evaluation method. The optimised output parameters are used to set up a complete system that is compared with a non-optimised one by designing an unbiased evaluation model.     

Modeling culture in intelligent virtual agents

This work addresses the challenge of creating virtual agents that are able to portray culturally appropriate behavior when interacting with other agents or humans. Because culture influences how people perceive their social reality it is important to have agent models that explicitly consider social elements, such as existing relational factors. We addressed this necessity by integrating culture into a novel model for simulating human social behavior. With this model, we operationalized a particular dimension of culture—individualism versus collectivism—within the context of an interactive narrative scenario that is part of an agent-based tool for intercultural training. Using this scenario we conducted a cross-cultural study in which participants from a collectivistic country (Portugal) were compared with participants from an individualistic country (the Netherlands) in the way they perceived and interacted with agents whose behavior was either individualistic or collectivistic, according to the configuration of the proposed model. In the obtained results, Portuguese subjects rated the collectivistic agents more positively than the Dutch but both countries had a similarly positive opinion about the individualistic agents. This experiment sheds new light on how people from different countries differ when assessing the social appropriateness of virtual agents, while also raising new research questions on this matter.     

Efficiently finding the optimum number of clusters in a dataset with a new hybrid differential evolution algorithm: DELA

Clustering algorithms, a fundamental base for data mining procedures and learning techniques, suffer from the lack of efficient methods for determining the optimal number of clusters to be found in an arbitrary dataset. The few methods existing in the literature always use some sort of evolutionary algorithm having a cluster validation index as its objective function. In this article, a new evolutionary algorithm, based on a hybrid model of global and local heuristic search, is proposed for the same task, and some experimentation is done with different datasets and indexes. Due to its design, independent of any clustering procedure, it is applicable to virtually any clustering method like the widely used \(k\)-means algorithm. Moreover, the use of non-parametric statistical tests over the experimental results, clearly show the proposed algorithm to be more efficient than other evolutionary algorithms currently used for the same task.     

Characterizing peers communities and dynamics in a P2P live streaming system

Despite the large number of works devoted to understand P2P live streaming applications, most of them put forth so far rely on characterizing the static view of these systems. In this work, we characterize the SopCast, one of the most important P2P live streaming applications. We focus on its dynamics behavior as well as on the community formation phenomena. Our results show that SopCast presents a low overlay topology diameter and low end-to-end shortest path. In fact, diameter is smaller than 6 hops in almost 90 % of the observation time. More than 96 % of peers’ end-to-end connections present only 3 hops. These values combined may lead to low latencies and a fast streaming diffusion. Second, we show that communities in SopCast are well defined by the streaming data exchange process. Moreover, the SopCast protocol does not group peers according to their Autonomous System. In fact, the probability that a community contains 50 % of its members belonging to the same AS (when we observe the largest AS of our experiments) is lower then 10 %. Peers exchange more data with partners belonging to the same community instead of peers inside the same AS. For the largest AS we have, less than 18 % of peer traffic has been exchanged with another AS partners. Finally, our analysis provides important information to support the future design of more efficient P2P live streaming systems and new protocols that exploit communities’ relationships.     

Assessment of the Impact of Stormwater Characteristics on Clogging in Stormwater Filters

Hydraulic conductivity of granular filter media and its evolution over time is a key design parameter for stormwater filtration and infiltration systems that are now widely used in management of polluted urban runoff. In fact, clogging of filter media is recognised as the main limiting factor of these stormwater treatment systems. This paper focuses on the effect of stormwater characteristics on the clogging of stormwater filters. Effect of five different operational regimes has been tested in this study of sediment concentration; pollutant concentrations; stormwater sediment size; loading rate and stormwater loading/dosing regime and compared with the Base case. For each operational condition, five column replicates were tested. Results suggest that sediment concentration in stormwater is a significant parameter affecting hydraulic and treatment performance, eventually affecting longevity of these stormwater treatment systems. Further, the size of sediments (and their relation to the size of filter media grains) in stormwater was found to be an important parameter to be considered in design of coarse filters with high infiltration rates that are used for stormwater treatment. As expected, the addition of metals and nutrients had limited or no contribution to changes in hydraulic or sediment removal performance of the studied stormwater filters. Whilst loading rate was found to be an important parameter affecting the hydraulic and treatment performance of these systems, any variation in the stormwater loading regime had a limited effect on their performance. This study therefore develops an understanding of the effect of catchment characteristics on design of filters and hence their longevity and maintenance needs.