Vladimir Estivill-Castro

• Doctor of Philosophy
• Professor (Full) at Pompeu Fabra University

Smart-cities are heavily linked to decision-making based on data. However, despite the different waves of open data, those pertaining to the citizens themselves are not directly contributing to citizen participation and engagement. This is particularly true of the fine grained data regarding utilities consumption despite the huge potential to impac...

The emergence of autonomous vehicles has raised ethical considerations regarding their controlling software. The focus is on defining ethical settings that determine the response to moral dilemmas, akin to the Trolley Problem. The study explores the interaction with a robot that engages users in a dialogue about such ethical dilemmas, allowing user...

Decomposition allows for managing complexity. We show that executable models of behaviour are significantly more decomposable when using a time-triggered semantics than an event-driven semantics. Therefore, we adopt logic-labelled finite machines LLFSMs and show that deterministic static schedules are derived to guarantee value-domain properties an...

Interactive machine learning (IML) enables models that incorporate human expertise because the human collaborates in the building of the learned model. Moreover, the expert driving the learning (human-in-the-loop-learning) can steer the learning objective, not only for accuracy, but perhaps for discrimination or characterisation rules, where isolat...

Interactive machine learning (IML) enables the incorporation of human expertise because the human participates in the construction of the learned model. Moreover, with human-in-the-loop machine learning (HITL-ML), the human experts drive the learning, and they can steer the learning objective not only for accuracy but perhaps for characterisation a...

We show the advantages of using Swift as the programming language for behaviours on the Pepper and Nao robots as used with the RoboCup Standard Platform League and the RoboCup@Home - Social Standard Platform. We show that Swift is not only incorporating modern features of object-oriented programming and functional programming, but is also now a sta...

The Internet of Things (IoT) has hugely expanded. There are more devices than people and device numbers are still growing exponentially. These IoT devices are becoming more important and integrated into our lives, and their sophisticated behaviour has prompted the term “Internet of Behaviours”. Model-Driven Software Engineering (MDSE) and especiall...

Matching lists of addresses is an increasingly common task executed by business and governments alike. However, due to security issues, this task cannot always be performed using cloud computing. Moreover, addresses can arrive with spelling errors that can cause non-matches or ‘false negatives’ to occur. Our proposed framework, PostMatch, provides...

WiFi is prevalent in mobile robotics and the Internet of Things (IoT) despite the fact that many applications in these environments re- quire real-time communication. WiFi does not guarantee timeliness and often introduces unbounded communication delays. We present TTWiFi, a time-triggered protocol for fault-tolerant real-time communication between...

As \emph{artificial intelligence} (AI) systems are increasingly involved in decisions affecting our lives, ensuring that automated decision-making is fair and ethical has become a top priority. Intuitively, we feel that akin to human decisions, judgments of artificial agents should necessarily be grounded in some moral principles. Yet a decision-ma...

A malicious data-miner can infer users' private information in online social networks (OSNs) by data mining the users' disclosed information. By exploring the public information about a target user (i.e. an individual or a group of OSN users whose privacy is under attack), attackers can prepare a training data set holding similar information about...

We present a new Decision Tree Classifier (DTC) induction algorithm that produces vastly more interpretable trees in many situations. These understandable trees are highly relevant for explainable artificial intelligence, fair automatic classification, and human-in-the-loop learning systems. Our method is an improvement over the Nested Cavities (NC...

***Note: this paper was awarded the best paper award at the SIMULTECH 2021 congress*** The exceptionally high virulence of COVID-19 and the patients' precondition seem to constitute primary factors in how pro-inflammatory cytokines production evolves during the course of an infection. We present a System Dynamics Model approach for simulating the p...

RoboCup is one of the major global AI events, gathering hundreds of teams from the world’s best universities to compete in various tasks ranging from soccer to home assistance and rescue. The commonality of these three seemingly dissimilar tasks is that in order to perform well, the robot needs to excel at the all major AI tasks: perception, contro...

Typically, humans interact with a humanoid robot with apprehension. This lack of trust can seriously affect the effectiveness of a team of robots and humans. We can create effective interactions that generate trust by augmenting robots with an explanation capability. The explanations provide justification and transparency to the robot’s decisions....

Modelling and simulation are techniques instrumental in the engineering and design of complex systems. The reason is that both these techniques can anticipate possible failures when corrections are less costly to incorporate. Nevertheless, a correct behaviour is no guarantee, especially with software systems and their ubiquitous modelling notation:...

The interest in interpretable models that are not only accurate but also understandable is rapidly increasing; often resulting in the machine-learning community turning to decision tree classifiers. Many techniques of growing decision trees use oblique rules to increase the accuracy of the tree and decrease its overall size, but this severely limit...

Online social network (OSN) users share various types of personal information with other users. By analysing such personal information, a malicious data miner (or an attacker) can infer the sensitive information about the user which has not been disclosed publicly. This is generally known as attribute inference attack. In this study, we propose a p...

Online social network (OSN) users share various types of personal information with other users. By analysing such personal information , a malicious data miner (or an attacker) can infer the sensitive information about the user which has not been disclosed publicly. This is generally known as attribute inference attack. In this study, we propose a...

We propose a new clustering technique called HeMI++. It uses cleansing and cloning operations that help to produce sensible clusters. HeMI++ learns necessary properties of a reasonable clustering solution for a dataset from a high-quality initial population, without requiring any user input. It then disqualifies the chromosomes that do not satisfy...

We introduce a cluster evaluation technique called Tree Index. Our Tree Index algorithm aims at describing the structural information of the clustering rather than the quantitative format of cluster-quality indexes (where the representation power of clustering is some cumulative error similar to vector quantization). Our Tree Index is finding margi...

To integrate robots into humans' environment, robots need to make their decision-making process transparent to increase humans' trust in robots. Explanations from a robot are a promising way to express "how" a decision is made and "why" the decision made is the best. We performed a user study investigating the effect of the explanations from a robo...

There is exciting news in recent developments suggesting the potential to treat some human cancers by stimulating the patients own immune system. However, there is still much to understand; therefore, modelling the battle between those cells that are constituents of the human immune system against tumorous cells can significantly provide insights a...

Models are used mainly to communicate among humans the most relevant aspects of the item being modelled. Moreover, for achieving impact in modern complex applications, modelling languages and tools must support some level of composition. Furthermore, executable models are the foundations of model-driven development; therefore, it is crucial that we...

We have developed three lessons supported by the principles of inquiry-based learning (IBL) and problem-based learning (PBL) in educational robotics with the aim of steering and emphasising the mathematics aspects of the curriculum and the role of mathematics in STEM, while also touching on the social context and impact of STEM. Our goal is to insp...

When humans interact with robots in daily life, each human has a different attitude towards robots that may directly affect human-robot trustworthy relationships. By attitude , we mean any mental disposition matured through experience. A negative attitude is the psychological factors that prevent a human from interacting with robots in daily life a...

Deep Learning has become exceptionally popular in the last few years due to its success in computer vision [1, 2, 3] and other fields of AI [4, 5, 6]. However, deep neural networks are computationally expensive, which limits their application in low power embedded systems, such as mobile robots. In this paper, an efficient neural network architectu...

Deep Learning has become exceptionally popular in the last few years due to its success in computer vision and other fields of AI. However, deep neural networks are computationally expensive, which limits their application in low power embedded systems, such as mobile robots. In this paper, an efficient neural network architecture is proposed for t...

Deep Learning has become exceptionally popular in the last few years due to its success in computer vision and other fields of AI. However, deep neural networks are computationally expensive, which limits their application in low power embedded systems, such as mobile robots. In this paper, an efficient neural network architecture is proposed for t...

Convolutional neural networks (CNNs) are the state-of-the-art method for most computer vision tasks. But, the deployment of CNNs on mobile or embedded platforms is challenging because of CNNs’ excessive computational requirements. We present an end-to-end neural network solution to scene understanding for robot soccer. We compose two key neural net...

To investigate, how a Robot’s communication ability in terms of explanations can cultivate better trust relationships between a Robot and it’s human teammates. We opted a partial information game-playing environment, to immerse interaction between humans and a Robotic Agent. We designed our Robotic Agent as a Knowledge-Based (KB) Robotic Agent that...

Millions of users place data about themselves on on-line social networks and, while probably they have an interest on some of this information to be publicly available, they certainly may consider some of this information shall remain confidential. Simultaneously, the data provides benefits as such data enables personalization which increases the q...

We consider the re-identification of users when they participate in several different on-line social networks, potentially using several different accounts. The re-identification of users serves several purposes: (i) commercial use so as to avoid redundant mailing to the same user; (ii) enhancement of the information available about these users by...

The inclusion of autonomous robots among everyday human environments has suggested that these robots will be facing ethical decisions regarding trade-offs where machines will choose some human attributes over the attributes of other humans. We argue in this paper that on a regular instance, algorithms for such decisions should not only be determini...

Studies in Information Retrieval and Technology Enhanced Learning have not been able yet to propose reliable support to students and teachers when seeking educational resources on the Web. The driving force of web-search has been to match the topic of a query with the topic of documents. This paper involves Natural Language Learning approaches for...

Online social networks (OSNs) are currently a popular platform for social interactions among people. Usually, OSN users upload various contents including personal information on their profiles. Not all information uploaded on OSNs are public. The ability to infer users' hidden information or information that has not been even uploaded (i.e. private...

We present an architecture that enables run-time verification with executable models of behaviour. Our uniform modelling paradigm is logic-labelled finite-state machines (LLFSMs). Behaviours are constructed by parameterizable, loadable, and suspendable LLFSMs executed in predictable sequential schedules, but they are also verified at run-time by LL...

A well-known method to produce accurate predictive models is to apply algorithms for feature selection and feature reduction. These algorithms describe an item with a subset of its attributes that is expected to be the smallest possible without compromising the actual representation of the object, and consequently the entire classification. However...

Knowledge discovery from data demands that it shall be the data themselves that reveal the groups (i.e. the data elements in each group) and the number of groups. For the ubiquitous task of clustering, K-MEANS is the most used algorithm applied in a broad range of areas to identify groups where intra-group distances are much smaller than inter-grou...

Abstract—Sensitive information of an Online Social Network (OSN) user can be discovered through sophisticated data mining, even if the user does not directly reveal such information. Malicious data miners can build a decision tree/forest from a data set having information about a huge number of OSN users and thereby learn general patterns which the...

Logic-Labeled Finite State Machines (LLFSMs) offer model-driven software development (MDSD) while enabling correctness at a high level due to their transparent semantics that enables testing as well as formal verification. This combination of the three elements (MDSD, validation, and verification) results in more reliable behaviour of software comp...

Online auctions have become an increasingly popular and convenient way for conducting ecommerce transactions on the Web. However, the rapid surge of users participating in online auctions has led to auction fraud. Among the types of auction fraud, the most prominent is Shill bidding. Shill bidding is intentionally fake bidding by a seller on his/he...

The possibility that an unauthorised agent is able to infer a user’s hidden information (an attribute’s value) is known as attribute inference risk. It is one of the privacy issues for Facebook users in recent times. An existing technique [1] provides privacy by suppressing users’ attribute values from their profiles. However, suppression of an att...

We describe how control/status OO-messages on shared-memory middleware can provide better performing control of a micro-air vehicle (MAV). To illustrate this, we provide a new hardware abstraction for a controller application that is completely analogous to the popular ardrone_autonomy (AA) package that enables the Parrot AR Drone 2.0 quadcopter to...

Many popular clustering techniques including K-means require various user inputs such as the number of clusters k, which can often be very difficult for a user to guess in advance. Moreover, existing techniques like K-means also have a tendency of getting stuck at local optima. As a result, various evolutionary algorithm based clustering techniques...

Validity is a fundamental aspect of any machine learning approach. All the three types of current validity approaches (external, internal, and relative) have serious drawbacks and are computationally expensive. This chapter discusses why there are so many proposals for clustering algorithms and why they detach from approaches to validity. It presen...

In the Technology Enhanced Learning (TEL) community, the problem of conducting reproducible evaluations of recommender systems is still open, due to the lack of exhaustive benchmarks. The few public datasets available in TEL have limitations, being mostly small and local. Recently, Massive Open Online Courses (MOOC) are attracting many studies in T...

We incorporate logic programs (in particular Prolog) into reactive systems. We do this using Logic-labeled finitestate machines (LLFSMs), whose non-event-driven nature results in a deterministic schedule. We also advocate the use of a middleware under the Pull-approach, as opposed to the, currently very common, Push-approach, achieving deterministi...

We say that there is a community structure in a graph when the nodes of the graph can be partitioned into groups (communities) such that each group is internally more densely connected than with the rest of the graph. However, the challenge seems to specify what is to be dense, and what is relatively more connected (there seems to exist a similar s...

The development of companion animal robots is of growing interest. These robots have recently been marketed to older adults with dementia as a means of encouraging social engagement and reducing behavioural and psychological symptoms of dementia. This paper outlines the results of a pilot study that sought to assess the feasibility and effect of us...

A malevolent data miner can use data mining techniques in order to learn confidential information of social networking site users that the users did not disclose ; and thereby the data miner can breach individual privacy of a social networking site user . However, the information items in a social network are not only the attributes of users but al...

The Minimum Vertex Cover (MinVC) problem is a well-known NP-hard problem. Recently there has been great interest in solving this problem on real-world massive graphs. For such graphs, local search is a promising approach to finding optimal or near-optimal solutions. In this paper we propose a local search algorithm that exploits reduction rules and...

Using the web for communication, purchases, searching information and/or socializing generates data, about ourselves, our connections and our activities, which is collected easily. In online social networks, users volunteer perhaps what is considered more personal information to their selected circles. But each person has personal preferences about...

Internet users in general and on-line social networks users in particular are becoming more savvy about masking data they consider private. However, some of this masked data may be inferable from other data the user has not masked. Furthermore, even if a user masks all its data, it may still be inferable from the unmasked data of its friends, due t...

Recent communication by Minghui Jiang has brought to my attention that I overlooked faults in the arguments built while collaborating closely with my PhD student Apichat Heednacram and his co-supervisor Francis Suraweera. These errors unfortunately also escaped the scrutiny of peer-reviews and the formal process of examination. Some results in Apic...

Millions of users place data about themselves on on-line social networks and, while probably they have an interest on some of this information to be publicly available, they certainly may consider some of this information shall remain confidential. Simultaneously, the data provides benefits as such data enables personalization which increases the q...

There are many areas where software components must interact witch each other and where middleware provides the appropriate benefits of robustness, decoupling, and modularisation. However, there is a potential performance overhead that, for autonomous robotic and embedded systems, may be critical. Proposals for robotic middleware continue to emerge...

We propose to test software models with software models. Model-Driven Software Development proposes that soft- ware is to be constructed by developing high-level models that directly execute or generate most of the code. On the other hand, Test-Driven development proposes to produce tests that validate the functionality of the code. This paper brin...

We present gusimplewhiteboard, a software architecture analogous to ROS:services and ROS: messages, that enables the construction and extremely efficient inter-process relaying of message-types as C++11 objects, All gusimplewhiteboard objects reside in shared memory. Moreover, our principle is to use idempotent message communication, in direct cont...

Formal methods have seen emergent success recently with the deployment of Event-B. However, Event-B explicitly postulates that models there are not executable. This seems to contradict the parallel emergence of model-driven development (MDD). We show here that logic-labeled finite-state machines (LLFSMs) are effective in carrying out the "correct f...

The standardization of planning problems by their descriptions in the PDDL has resulted in clear benchmarking of planners, and thus, in significant advances in reliable and efficient planning packages. The output of these classical planners is a plan as sequence of actions for the controllable robots in the environment. We show here that, provided...

Logic-labeled finite-state machines are a formal mechanisms to represent behavior. These models have several advantages over event-driven finite-state machines. They have a formal semantics that enables model-checking; that is, formal verification. More importantly, they can be executed concurrently and produce simple behaviors for embedded systems...

We investigate the current state of the art in terms of formally verifiable protocols to form a formation of robots under unreliable communi- cation. We show that in practical terms, it is possible to obtain more efficient protocols with more appealing properties.

We propose a contrasting approach to the main stream direction that UML and STATEMATE have recently taken when using finite-state machines (FSMs) to model behaviour. That is, rather than the event-driven model that is currently dominant, we suggest to adopt a model of time, a synchronous model. We do support concurrency in our arrangements of FSMs...

Model-driven development results in directly runnable implementations, and therefore it is of utmost importance to formally verify and validate such models. However, model-checking usually faces the challenge of concurrent modules generating a state space equal to the Cartesian product of the state spaces of all modules. This is even more dramatic...

We propose vectors of finite-state machines whose transitions are labeled by formulas of a common-sense logic as the modeling tool for embedded systems software. We have previously shown that this methodology is very efficient in producing succinct and clear models (e.g., in contrast to plain finite-state machines, Petri nets, or Behavior Trees). W...

Simulation of models that specify behaviour of software in robots, embedded systems, and safety critical systems is crucial to ensure correctness. This is particularly important in conjunction with model-driven development, which is highly prevalent due to its numerous benefits. We use vectors of finite-state machines (FSMs) as our modelling tool....

A very successful tool for model-driven engineering of embedded systems is finite-state machines whose transitions are labeled with expressions of a common-sense logic. The deployment of models to different platforms and different programming languages makes it more imperative to confirm that the models are correct. However, systems are usually com...

Interpretability and usability of clustering results are of fundamental importance. A linear time method for transforming point clusters into polygons is explored. This method automatically translates a point data layer into a space filling layer where clusters are identified as some of the resulting regions. The method is based on robustly identif...

In game theory, a strategy for a player is dominant if, regardless of what any other player does, the strategy earns a better payoff than any other. If the payoff is strictly better, the strategy is named strictly dominant, but if it is simply not worse, then it is called weakly dominant. We investigate the parameterized complexity of two problems...

Model-Driven Development (MDD) has proven to be a very powerful tool to produce software for embedded systems that control sophisticated equipment. It is therefore even more critical that such software be verified to be correct and to clearly understand what the safety implications of potential failures in sensors, actuators or faults of the softwa...

We propose to model the behaviour of embedded systems by finite state machines whose transitions are modelled by predicates of non-monotonic logics. We argue that this enables modelling the behaviour in close parallelism to the requirements. Such requirements engineering also results in direct and automatic translation to implementation, minimising...