Self-learning abilities in autonomous systems are essential to improve their situational awareness and detection of normal/abnormal situations. In this work, we propose a graph matching technique for activity detection in autonomous agents by using the Gromov-Wasserstein framework. A clustering approach is used to discretise continuous agents' states related to a specific task into a set of nodes with similar objectives. Additionally, a probabilistic transition matrix between nodes is used as edges weights to build a graph. In this paper, we extract an abnormal area based on a sub-graph that encodes the differences between coupled of activities. Such sub-graph is obtained by applying a threshold on the optimal transport matrix, which is obtained through the graph matching procedure. The obtained results are evaluated through experiments performed by a robot in a simulated environment and by a real autonomous vehicle moving within a University Campus.

In this paper, we use preference learning techniques to model players’ emotional preferences in an AR mobile game. This exploratory study uses player behaviour to make these preference predictions. The described techniques successfully predict players’ frustration and challenge levels with high accuracy while all other preferences tested (boredom, excitement and fun) perform better than random chance. This paper describes the AR treasure hunt game we developed, the user study conducted to collect player preference data, analysis performed, and preference learning techniques applied to model this data. This work is motivated to personalize players’ experiences by using these computational models to optimize content creation and game balancing systems in these environments. The generality of our technique, limitations, and usability as a tool for personalization of AR mobile games is discussed.

This paper presents, a system capable of detecting unusual activities in crowds from real-world data captured from multiple sensors. The detection is achieved by classifying the distinct movements of people in crowds, and those patterns can be different and can be classified as normal and abnormal activities. Statistical features are extracted from the dataset collected by applying sliding time window operations. A model for classifying movements is trained by using Random Forest technique. The system was tested by using two datasets collected from mobile phones during social events gathering. Results show that mobile data can be used to detect anomalies in crowds as an alternative to video sensors with significant performances. Our approach is the first to detect any unusual behaviour in crowd with non-visual data, which is simple to train and easy to deploy. We also present our dataset for public research as there is no such dataset available to perform experiments on crowds for detecting unusual behaviours.

Believability is a hard concept to define in video games. It depends on how and what one considers to be "believable", which is often very subjective. In previous years, several researchers have tried to find ways of assessing such concepts in games through Turing Tests on agents, which were programmed to behave like a human instead of focusing only on winning. Examples are the Mario AI Competition and the 2K BotPrize. Given the small pool of explored parameters and a focus on programming the bots rather than the assessment, in this paper we present work examining different methods of evaluating believability in video games. We explore believability through recorded gameplay and allow judges to analyze it. However, we use different parameters - such as ranking rather than binary answers - for asking how human-like the presented behaviours are. The objective of this study is to analyze the different ways believability can be assessed, for humans and non-player characters (NPCs) by comparing how results between them and scores are affected in both when changing the parameters. In order to provide a more general analysis, the study is carried out using two different racing games rather than one. Results show that these parameters have indeed changed the overall results of the study and how important it is to be able to generalize these concepts in game AI, given how clear it is that believability is dependent on genre, game and even the design of the questionnaire.

The role of affective states in cognitive performance has long been an area of interest in cognitive science. Recent research in game-based cognitive training suggest that cognitive games should incorporate real-time adaptive mechanisms. These adaptive mechanisms would change the game's difficulty according to the player's performance in order to provide appropriate challenges and thus, achieve a real cognitive improvement. However, these mechanisms currently ignore the effects of valence and arousal on the player's cognitive skills. In this paper we investigate how working memory (WM) performance is affected when playing a VR game, and the effects of valence and arousal in this context. To this aim, a custom video game was created for Desktop and VR. Three difficulty levels were designed to evoke different levels of arousal while maintaining the same memory load for each difficulty level. We found an improvement in WM performance when playing in VR compared to Desktop. This effect was particularly pronounced in those with a low WM capacity. Significantly higher levels of valence and arousal were self-reported when playing in VR. We explore the impact that reported affective states could have in the player's WM performance. We suggest that high levels of arousal and positive valence can lead players to a flow state [1] that may have a positive impact on the player's WM performance.

Researchers from different disciplines have examined crowd behavior in the past by employing a variety of methods including ethnographic studies, computer vision techniques and manual annotation based data analysis. However, because of the inherent difficulties in collecting, processing and analyzing the data, it is difficult to obtain large data sets for study. In this work we present a system for detecting stationary interactions inside crowds, depending entirely on the sensors available in a modern smartphone device such as Bluetooth Smart (BLE) and Accelerometer. By utilizing Apple's iBeaconTM implementation of Bluetooth Smart using SensingKit1, our open-source multi-platform mobile sensing framework [1], we are able to detect the proximity of users carrying a smartphone in their pocket. We then use an algorithm based on graph theory to predict group interactions inside the crowd. Previous work in this area has been limited to the detection of interactions between only two people and therefore our approach goes beyond current state of the art in its ability to detect group formations with more than two people involved. Our approach is particularly beneficial to the design and implementation of crowd behavior analytics, design of influence strategies, and algorithms for crowd reconfiguration.

Today's smartphones come equipped with a range of advanced sensors capable of sensing motion, orientation, audio as well as environmental data with high accuracy. With the existence of application distribution channels such as the Apple App Store and the Google Play Store, researchers can distribute applications and collect large scale data in ways that previously were not possible. Motivated by the lack of a universal, multi-platform sensing library, in this work we present the design and implementation of SensingKit, an open-source continuous sensing system that supports both iOS and Android mobile devices. One of the unique features of SensingKit is the support of the latest beacon technologies based on Bluetooth Smart (BLE), such as iBeacon and Eddystone. We evaluate and compare the power consumption of each supported sensor individually, using an iPhone 5S device running on iOS 9. We believe that this platform will be beneficial to all researchers and developers who plan to use mobile sensing technology in large-scale experiments.

A variety of event detection algorithms for microblog services have been proposed, but their accuracy relies on the microblog feeds they analyse. Existing research explores datasets that are collected using either a set of manually predefined terms or information from external sources. These methods fail to provide comprehensive and quality feeds for real-time event detection. In this paper, we present a novel adaptive keyword identification approach to retrieve a greater amount of event relevant content. This approach continuously monitors emerging hashtags and rates them by their similarity to specific pre-defined event hashtags using TF-IDF vectors. Top rated emerging hashtags are added as filter criteria in real time. By comparing our proposed approach, called CETRe (Content-based Event Tweet Retrieval) with an existing baseline approach applied to real-world events, we show that CETRe not only identifies event topics and contents, but also enables better event detection.

With the rapid rise in variety of available smartphones today and their rich sensing capabilities, there is an increasing interest in using mobile sensing in large-scale experiments and commercial applications. Motivated by the lack of a universal, multi-platform library, in this paper we present SensingKit, an efficient, open-source, client-server system that supports both iOS and Android mobile devices. SensingKit is capable of continuous sensing the device's motion (Accelerometer, Gyroscope, Magnetometer), location (GPS) and proximity to other smartphones (Bluetooth Smart). The data are temporarily saved to the device's memory and transmitted to a server for further analysis over any Internet connection. We believe that this platform will be beneficial to all researchers and developers who need to perform mobile sensing in their applications and experiments.

Researchers have capitalized on microblogging services, such as Twitter, for detecting and monitoring real world events. Existing approaches have based their conclusions on data collected by monitoring a set of pre-defined keywords. In this paper, we show that this manner of data collection risks losing a significant amount of relevant information. We then propose an adaptive crawling model that detects emerging popular hashtags, and monitors them to retrieve greater amounts of highly associated data for events of interest. The proposed model analyzes the traffic patterns of the hashtags collected from the live stream to update subsequent collection queries. To evaluate this adaptive crawling model, we apply it to a dataset collected during the 2012 London Olympic Games. Our analysis shows that adaptive crawling based on the proposed Refined Keyword Adaptation algorithm collects a more comprehensive dataset than pre-defined keyword crawling, while only introducing a minimum amount of noise.

A probabilistic approach for outdoor location estimation using GSM received signal strength (RSS) from base stations (BSs) is presented. The proposed approach first divides the region of interest into different clusters based on deviations from the path loss model for each RSS component. In each cluster, the proposed algorithm uses principal component analysis (PCA) to intelligently transform RSS into new uncorrelated dimensions. This retains accuracy by not losing the substantial RSS correlations in each cluster, but also accommodates the different RSS distributions in each cluster. Our experiments are conducted in a real GSM outdoor environment. The proposed approach is compared with a traditional probabilistic algorithm for three different area partitioning methods. The experimental results show that the positioning accuracy is significantly improved and our clustering scheme gives good support for location estimation. Furthermore, it also can be concluded that the clustering scheme created by using deviation RSS based on Mahalanobis distance performs better than that using deviation based on Euclidean distance in a complex environment. What's more, the proposed method can reduce the number of training data used while maintaining the accuracy required.

This paper presents results for an approach for indoor location estimation that integrates received signal strength (RSS) data from both WiFi and GSM networks. Previous work has focused on relatively small indoor environments. In many potential applications, getting approximate location information, such as in which room the mobile user is, is adequate. A hierarchical clustering method is used to partition the RSS space. To choose the best transmitters in a partition, we assess the amount of RSS variance that is attributable to different base stations (BSs) or access points (APs) by transforming the RSS tuples into principal components (PCs). This allows us to retain most of the useful information of detectable transmitters in fewer dimensions. In our experiments, we collected WiFi and cellular RSS on the 2nd and 3rd-floor electronic engineering (EE) building in Queen Mary campus. The experiment results show that the proposed method can provide a good accuracy of room prediction, especially when we integrate WiFi RSS with GSM RSS together to do the positioning.

We analyse different worm and patch propagation models along with the ones we have developed and evaluated as a part of our ongoing passive P2P worm & patch modelling project. This is followed by a brief discussion on worm detection mechanisms proposed by various authors. Towards the very end of this article, we propose a distributed framework for passive worm throttling in P2P networks and discuss its feasibility and efficiency keeping in view different design considerations.

Peer-to-Peer (P2P) networks are becoming very popular these days. The scattered environment in P2P networks makes it different from the traditional networks. Each member in P2P network can act as a client as well as a server, which is different from its traditional counterpart such as Internet, where a central server is required to control the network. Finding resources in P2P systems is a major issue. Extensive search may be needed to resolve queries in the network. In traditional P2P search algorithms such as in Gnutella protocol V0.4 search is via query flooding with significant exploration cost. In this paper, a group-structured P2P network is proposed. The group-based search provides well-organized searching technique by targeting particular group or community. We propose a new protocol for building and repairing of overlay topologies based on the formation of interest-based superpeers. An interest-based superpeer algorithm creates groups or societies that have common interests. Simulations indicate that the proposed protocol is highly capable and powerful even in the dynamic nature of P2P networks, where nodes are continually joining and leaving the network and the protocol supports restoration of the network even after the catastrophic failure of interest-based superpeers.

A community live action game over Bluetooth, namely Blue Danger, is proposed in this paper. The market penetration of Bluetooth provides an excellent media for such game in which a large number of players is required. Bluetooth has a relatively short transmission range that allows a dynamic mobile ad hoc network environment to be created which is ideal for the action game proposed. In addition, reliability of Bluetooth as well as its detection and tracking have been examined during the development of the game. A prototype game framework has been built and tested.