Andrew Koster

Andrew Koster
eDreams Odigeo · Data Science

PhD in Computer Science

About

39
Publications
13,762
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
216
Citations
Introduction
Andrew Koster currently works at the Artificial Intelligence Research Institute, Spanish National Research Council (IIIA-CSIC). Andrew researches topics in Data Mining, Distributed Computing and Artificial Intelligence. His current project is 'Socio-cognitive Simulation for Advancing Digital Education'.
Additional affiliations
September 2016 - present
Spanish National Research Council
Position
  • Researcher
January 2014 - May 2016
Samsung
Position
  • Researcher
Description
  • Apply AI technologies to solve social problems.
August 2012 - December 2013
Universidade Federal do Rio Grande do Sul
Position
  • PostDoc Position

Publications

Publications (39)
Article
Full-text available
This paper presents a non-prioritized belief change operator, designed specifically for incorporating new information from many heterogeneous sources in an uncertain environment. We take into account that sources may be untrustworthy and provide a principled method for dealing with the reception of contradictory information. We specify a novel Data...
Book
Full-text available
This book constitutes the refereed proceedings of the First International Workshop, SOCIALEDU 2015, held in Stanford, CA, USA, in August 2015. The workshop aimed to discuss computational models of social computing applied to Digital Education. The 9 revised full papers presented were carefully reviewed and selected from 12 submissions. The papers a...
Article
This survey is the first to review the combination of computational trust and argumentation. The combination of the two approaches seems like a natural match, with the two areas tackling different aspects of reasoning in an uncertain, social environment. We discuss the different areas of research and describe the approaches taken so far, analysing...
Article
Full-text available
Trust models as thus far described in the literature can be seen as a monolithic structure: a trust model is provided with a variety of inputs and the model performs calculations, resulting in a trust evaluation as output. The agent has no direct method of adapting its trust model to its needs in a given context. In this article, we propose a first...
Article
In open multi-agent systems trust models are an important tool for agents to achieve effective interactions. However, in these kinds of open systems, the agents do not necessarily use the same, or even similar, trust models, leading to semantic differences between trust evaluations in the different agents. Hence, to successfully use communicated tr...
Book
This book constitutes the refereed post-conference proceedings of the First International Workshop on Artificial Intelligence in Health, AIH 2018, in Stockholm, Sweden, in July 2018. This workshop consolidated the workshops CARE, KRH4C and AI4HC into a single event. The 18 revised full papers included in this volume were carefully selected from th...
Conference Paper
The User Experience (UX) of a software product is influenced by pragmatic and hedonic aspects, and it is necessary to choose a UX evaluation method that takes both of these aspects into account. In this paper, we report on the lessons learned from applying different UX evaluation methods (3E, 3E*, SAM, MAX, EM, Think Aloud, and Observation) in prot...
Conference Paper
Full-text available
This work discusses the usability experiments conducted around a proof-of-concept implementation of a novel tablet-based Digital Teaching Platform (DTP). The platform is intended to address specific issues with tablet usage in a classroom setting, and address problems with technology adoption in education, particularly in Brazil. We evaluated the D...
Conference Paper
This work seeks to build a new kind of classroom experience by rethinking how educational content is currently transmitted and consumed at schools. This work presents the results of applying the Short Bridge Method in the education context. We evaluate how this approach contributes to the class composition process by providing tools that support ed...
Conference Paper
Full-text available
Agent-BasedModellingandSimulation(ABMS)isaresearch methodology for studying complex systems that has been used with suc- cess in many social sciences. However, it has so far not been applied in education research. We describe some of the challenges for applying ABMS in the area of education, and discuss some of the potential ben- efits. We describe...
Conference Paper
Full-text available
This paper presents an empirical analysis of an automated measure for in-classroom student engagement. The paper presents (1) a novel learning metric to measure student engagement and (2) a curated data set, collected with a novel tool for logging usage data from a tablet-based digital teaching platform. We show that in this data, the student engag...
Data
This dataset is made available under the Open Data Commons Attribution License: http://opendatacommons.org/licenses/by/1.0/. - See more at: http://opendatacommons.org/licenses/by/ Please reference the following manuscript for attribution: Koster, Andrew, Tiago Primo, Allysson Oliveira and Fernando Koch. Toward Measuring Student Engagement: A Data...
Data
This dataset is made available under the Open Data Commons Attribution License: http://opendatacommons.org/licenses/by/1.0/. - See more at: http://opendatacommons.org/licenses/by/ Please reference the following manuscript for attribution: Koster, Andrew, Tiago Primo, Allysson Oliveira and Fernando Koch. Toward Measuring Student Engagement: A Data...
Book
This book constitutes the refereed proceedings of the 7th International Workshop on Collaborative Agents Research and Development, CARE 2016, held in Singapore in May 2016 and Second International Workshop on Social Computing in Digital Education, SocialEdu 2016, held in Zagreb, Croatia, in June 2016. For CARE 2016 there were 4 papers selected out...
Conference Paper
Full-text available
This paper describes an experimental evaluation of the main machine learning supervised techniques to be used for the human activities recognition in the context of technological education using data collected from smartphones sensors. The overall goal is to use the recognition of activities to identify students with attention deficit or hyperac-ti...
Conference Paper
Full-text available
We introduce innovations in a Digital Teaching Platform (DTP) through tools centred on supporting the teacher. We focus on the utilisation of data about the students and the class in order to recommend actions and content for the teacher. For this, we need a platform with novel capabilities. First, we augment the content delivery application with d...
Article
Full-text available
The problem of finding parking slots imposes both societal and infrastructural issues in modern cities. It is a daily hurdle that affects millions of people, but existing approaches fail to solve this conundrum. Thus, there is an urgent demand for reputable, motivated, and replicable solutions that can be used by cities of any size. We are proposin...
Conference Paper
Full-text available
Context-awareness is an essential requirement in crafting recommender systems that provide serendipity, i.e. “pleasant surprises”, independently of human command. These solutions must be able to infer interactions based on data from sensors and recognised activities in order to infer what is useful information and when to deliver it. For that, we a...
Conference Paper
Full-text available
Current solutions to recommend available parking spaces rely on options like: intentional user feedback; installing data collectors in volunteering fleet vehicles, or; installing static sensors to monitor available parking spaces. In this paper we propose a solution based application that runs on commodity smartphones and makes use of the advanced...
Patent
Full-text available
A presente invenção se refere a um método para auto adaptar um classificador de movimentos (assinatura) de estacionamento ao estilo de um usuário individual. As assinaturas do movimento, como detectadas pelo Aplicativo Recomendador de Vagas de Estacionamento 102, 301, 501) são enviadas a um recurso de tecnologia da informação remoto (i.e. sistema e...
Chapter
Full-text available
In this chapter we discuss the problem of communicating about trust and how semantic technologies can help. We briefly introduce these semantic technologies and then discuss two well-known ontologies of trust: \({\mathcal{L}}_{Rep}\) and FORe. However, defining a shared language for trust ignores the personal and subjective aspect of trust, which a...
Conference Paper
Full-text available
VANETs allow for unprecedented amounts of information to be sent between participants in traffic. Unfortunately, without countermeasures, they also allow selfish agents to take advantage of communication to improve their own utility. In this paper we present a novel framework for dealing with potentially untrustworthy information. The framework con...
Chapter
Full-text available
In this paper we analyse the trade-off between privacy-preservation methods and the quality of data mining applications, within the specific context of the smart grid. The use of smart meters to automate data collection is set to solve the problem of electricity theft, which is a serious concern in developing nations. Nevertheless, the unlimited us...
Conference Paper
Full-text available
When people need help with day-to-day tasks they turn to family, friends or neighbours to help them out. Despite an increasingly networked world, technology falls short in supporting such daily tasks. u-Help provides a platform for building a community of helpful people and supports them in finding volunteers for day-to-day tasks. It relies on thre...
Chapter
Full-text available
In this chapter we discuss the ways in which trust can be combined with argumentation. This is a new field of research that is showing promising approaches to a number of problems in both argumentation and trust. We discuss three ways in which trust and argumentation are combined. The first is to use the trustworthiness of an agent as a level of co...
Conference Paper
Full-text available
Agents in open multi-agent systems must deal with the difficult problem of selecting interaction partners in the face of uncertainty about their behaviour. This is especially problematic if they have to interact with an agent they have not interacted with before. In this case they can turn to their peers for information about this potential partner...
Conference Paper
Full-text available
In open multi-agent systems trust is necessary to improve cooperation by enabling agents to choose good partners. Most trust models work by taking, in addition to direct experiences, other agents' communicated evaluations into account. However, in an open multi-agent system other agents may use different trust models and as such the evaluations the...
Conference Paper
Full-text available
In heterogeneous multi-agent systems trust is necessary to improve interactions by enabling agents to choose good partners. Most trust models work by taking, in addition to direct experiences, other agents' communicated evaluations into account. However, in an open MAS other agents may use different trust models and the evaluations they communicate...
Conference Paper
Full-text available
In open multi-agent systems trust models are an important tool for agents to achieve effective interactions. However, the agents do not necessarily use similar trust models, leading to semantic differences between trust evaluations in the different agents. We show how to form a trust alignment by considering the interactions agents share. We descri...
Conference Paper
Full-text available
We present a mathematical framework for communicating about trust in terms of interactions. We argue that sharing an ontology about trust is not enough and that interactions are the building blocks that all trust-and reputation models use to form their evaluations. Thus, a way of talking about these interactions is essential to gossiping in open he...
Article
Full-text available
Knowing which agents to trust is an important problem in open multi-agent systems. A way to help solve this prob-lem is by allowing agents to relay information about trust to each other. We argue trust is a subjective phenomenon and therefore needs aligning. We present a mathematical framework for communicating about trust in terms of inter-actions...
Article
Full-text available
The Autonomous Intelligent Robot (AIR) Laboratory consists of researchers and students aiming to develop goal-directed, adaptive, and autonomous behaviors in a wide variety of robots. Since the foundation of the Lab in 1997, we have done a number of different projects including (1) Developing self-localisation algorithms and behaviors for the Pione...
Article
Full-text available
In open multi-agent systems trust models are an important tool for agents to achieve effective interactions. However, in these kinds of open systems, the agents do not necessarily use the same, or even similar, trust models, leading to semantic differences between trust eval-uations in the different agents. Hence, to successfully use communicated t...
Article
Full-text available
We present a mathematical framework and an implemen-tation of a proof of concept for communicating about trust in terms of interactions. We argue that sharing an ontology about trust is not enough and that interactions are the building blocks that all trust-and reputation models use to form their evaluations. Thus, a way of talking about these inte...
Article
Full-text available
In this position paper we explain why the alignment of trust for computational agents is a problem which requires closer consideration than it has previously been given. We give a review of related work from various fields of research and propose a general framework in which a solution for the alignment of trust should be found.
Article
Full-text available
The potential of pervasive mobile systems lies in the integra- tion of applications and services available on mobile devices. These services must be built on the premises of context- awareness, such that the application takes advantage of en- vironmental information to improve its performance. The challenge is to process more environmental informat...
Article
Full-text available
Using advanced pathfinding algorithms and a decision sys-tem based on utility-models for targets, we create agents that can largely act autonomously in a dynamic crisis scenario. We will show that using a clustering technique and the use of a new developed pathcost estimation algorithm the process of decisionmaking is simplified. The information gi...
Article
Full-text available
The Autonomous Intelligent Robot (AIR) Laboratory consists of researchers and students aiming to develop goal-directed, adaptive, and autonomous behaviors in a wide variety of robots. Since the foundation of the Lab in 1997, we have done a number of different projects including (1) Developing self-localisation algorithms and behaviors for the Pione...

Questions

Question (1)
Question
I am trying to use the HAR dataset (see attached link) to test my activity recognition algorithm. However, before using it to test my own method, I am trying to make sense of the data, and understand the features that are extracted.
As a simple test I tried to reproduce them, starting with the mean. I calculated the arithmetic mean of the first row of values in the data/train/Inertial Signals/body_acc_x_train.txt file. If I understand the explanation correctly, this should be the first value of the first line of data/train/X_train.txt However when computing the mean, I obtained 0.00226869, whereas the value in the X_train.txt file is 0.28858
This same discrepancy occurs for the y and z values. If I omit the division by 128 (number of samples in a window) then the value is nearer (at least of the same order of magnitude), but still further off than floating point errors should account for (just to be sure, I used the bigfloat package in my Python code with a precision of 15 to ensure the rounding errors were not the problem on my side.
I understand this is a rather niche question and sent it to the admin of the data set, unfortunately, he's out of office until the end of August, so thought I'd ask here in case someone has experience with using this dataset.

Network

Cited By

Projects

Projects (2)
Archived project
This project aimed to research and develop an end-to-end reference model to implement Digital Education for mobile computing in the classroom environment. This development revolved around technologies for Ambient Intelligence, Human-centred IoT, Machine Learning, Human Behaviour Analysis, and Recommendation Systems. The research aimed to join these technologies to advance the state-of-the-art in digital education towards: better understanding the educational environment in the classroom, reduce the working time of the teacher and provide information of decision support and recommendation of adjustments for continuous improvement of educational performance. The solution is involved of the following R&D lines: (i) Intelligent composition, with AI models to facilitate the process of creating advance digital content for education; (ii) Intelligent educational material, with content delivery application, instrumentation for capturing and recording user-content interactions, data analysis methods and content self-tuning for personalisation and context awareness; (iii) Models of multimodal learning. The project resulted in prototypes and field tests of innovative models of data analysis, content recommendation and activity recommendation. Moreover, this project promotes the ecosystem of developers and integrators around Samsung technology, directly contributing to the formation of Industry and technology.
Project
This project will research new agent-based modelling and simulation (ABMS) methods to model the socio-cognitive system in a classroom. We will: (i) evaluate existing learning analytics approaches for their suitability as part of a socio-cognitive model; (ii) iteratively design, calibrate and validate a socio-cognitive ABMS of a classroom; (iii) design and verify the usability and effectiveness of feedback mechanisms based on the output of the ABMS; and (iv) refine the socio-cognitive model to be aware of the macro features that the feedback loops introduce into the system.