A Survey of Active Learning Algorithms for Supervised Remote Sensing Image Classification

Defining an efficient training set is one of the most delicate phases for the success of remote sensing image classification routines. The complexity of the problem, the limited temporal and financial resources, as well as the high intraclass variance can make an algorithm fail if it is trained with a suboptimal dataset. Active learning aims at building efficient training sets by iteratively improving the model performance through sampling. A user-defined heuristic ranks the unlabeled pixels according to a function of the uncertainty of their class membership and then the user is asked to provide labels for the most uncertain pixels. This paper reviews and tests the main families of active learning algorithms: committee, large margin, and posterior probability-based. For each of them, the most recent advances in the remote sensing community are discussed and some heuristics are detailed and tested. Several challenging remote sensing scenarios are considered, including very high spatial resolution and hyperspectral image classification. Finally, guidelines for choosing the good architecture are provided for new and/or unexperienced user.

Innovation and new business through mutually beneficial collaboration and proactive procurement

ABSTRACT This dissertation focuses on new technology commercialization, innovation and new business development. Industry-based novel technology may achieve commercialization through its transfer to a large research laboratory acting as a lead user and technical partner, and providing the new technology with complementary assets and meaningful initial use in social practice. The research lab benefits from the new technology and innovation through major performance improvements and cost savings. Such mutually beneficial collaboration between the lab and the firm does not require any additional administrative efforts or funds from the lab, yet requires openness to technologies and partner companies that may not be previously known to the lab- Labs achieve the benefits by applying a proactive procurement model that promotes active pre-tender search of new technologies and pre-tender testing and piloting of these technological options. The collaboration works best when based on the development needs of both parties. This means that first of all the lab has significant engineering activity with well-defined technological needs and second, that the firm has advanced prototype technology yet needs further testing, piloting and the initial market and references to achieve the market breakthrough. The empirical evidence of the dissertation is based on a longitudinal multiple-case study with the European Laboratory for Particle Physics. The key theoretical contribution of this study is that large research labs, including basic research, play an important role in product and business development toward the end, rather than front-end, of the innovation process. This also implies that product-orientation and business-orientation can contribute to basic re-search. The study provides practical managerial and policy guidelines on how to initiate and manage mutually beneficial lab-industry collaboration and proactive procurement.

Forensic intelligence framework. Part II: study of the main generic building blocks and challenges through the examples of illicit drugs and false identity documents monitoring

The development of forensic intelligence relies on the expression of suitable models that better represent the contribution of forensic intelligence in relation to the criminal justice system, policing and security. Such models assist in comparing and evaluating methods and new technologies, provide transparency and foster the development of new applications. Interestingly, strong similarities between two separate projects focusing on specific forensic science areas were recently observed. These observations have led to the induction of a general model (Part I) that could guide the use of any forensic science case data in an intelligence perspective. The present article builds upon this general approach by focusing on decisional and organisational issues. The article investigates the comparison process and evaluation system that lay at the heart of the forensic intelligence framework, advocating scientific decision criteria and a structured but flexible and dynamic architecture. These building blocks are crucial and clearly lay within the expertise of forensic scientists. However, it is only part of the problem. Forensic intelligence includes other blocks with their respective interactions, decision points and tensions (e.g. regarding how to guide detection and how to integrate forensic information with other information). Formalising these blocks identifies many questions and potential answers. Addressing these questions is essential for the progress of the discipline. Such a process requires clarifying the role and place of the forensic scientist within the whole process and their relationship to other stakeholders.

Note méthodologique sur la collaboration entre praticiens dans le secteur ambulatoire. [Methodologic note on the collaboration between practitioners in the ambulatory care sector].

There are still few studies about the collaboration between ambulatory practitioners (physicians and paramedical services). Nevertheless, the interest seems to be growing for this aspect of health care; it involves indeed basic organisational problems as well as fundamental questions about quality of care and its economic implication. A basic problem is rooted in the inevitable contradiction of the efforts towards a highly qualified-i.e. specialised-care on the one hand and those towards continuity on the other hand.

Modeling the forensic two-trace problem with Bayesian networks

The forensic two-trace problem is a perplexing inference problem introduced by Evett (J Forensic Sci Soc 27:375-381, 1987). Different possible ways of wording the competing pair of propositions (i.e., one proposition advanced by the prosecution and one proposition advanced by the defence) led to different quantifications of the value of the evidence (Meester and Sjerps in Biometrics 59:727-732, 2003). Here, we re-examine this scenario with the aim of clarifying the interrelationships that exist between the different solutions, and in this way, produce a global vision of the problem. We propose to investigate the different expressions for evaluating the value of the evidence by using a graphical approach, i.e. Bayesian networks, to model the rationale behind each of the proposed solutions and the assumptions made on the unknown parameters in this problem.

Advanced Kernel Methods For Remote Sensing Image Classification

Résumé Suite aux recentes avancées technologiques, les archives d'images digitales ont connu une croissance qualitative et quantitative sans précédent. Malgré les énormes possibilités qu'elles offrent, ces avancées posent de nouvelles questions quant au traitement des masses de données saisies. Cette question est à la base de cette Thèse: les problèmes de traitement d'information digitale à très haute résolution spatiale et/ou spectrale y sont considérés en recourant à des approches d'apprentissage statistique, les méthodes à noyau. Cette Thèse étudie des problèmes de classification d'images, c'est à dire de catégorisation de pixels en un nombre réduit de classes refletant les propriétés spectrales et contextuelles des objets qu'elles représentent. L'accent est mis sur l'efficience des algorithmes, ainsi que sur leur simplicité, de manière à augmenter leur potentiel d'implementation pour les utilisateurs. De plus, le défi de cette Thèse est de rester proche des problèmes concrets des utilisateurs d'images satellite sans pour autant perdre de vue l'intéret des méthodes proposées pour le milieu du machine learning dont elles sont issues. En ce sens, ce travail joue la carte de la transdisciplinarité en maintenant un lien fort entre les deux sciences dans tous les développements proposés. Quatre modèles sont proposés: le premier répond au problème de la haute dimensionalité et de la redondance des données par un modèle optimisant les performances en classification en s'adaptant aux particularités de l'image. Ceci est rendu possible par un système de ranking des variables (les bandes) qui est optimisé en même temps que le modèle de base: ce faisant, seules les variables importantes pour résoudre le problème sont utilisées par le classifieur. Le manque d'information étiquétée et l'incertitude quant à sa pertinence pour le problème sont à la source des deux modèles suivants, basés respectivement sur l'apprentissage actif et les méthodes semi-supervisées: le premier permet d'améliorer la qualité d'un ensemble d'entraînement par interaction directe entre l'utilisateur et la machine, alors que le deuxième utilise les pixels non étiquetés pour améliorer la description des données disponibles et la robustesse du modèle. Enfin, le dernier modèle proposé considère la question plus théorique de la structure entre les outputs: l'intègration de cette source d'information, jusqu'à présent jamais considérée en télédétection, ouvre des nouveaux défis de recherche. Advanced kernel methods for remote sensing image classification Devis Tuia Institut de Géomatique et d'Analyse du Risque September 2009 Abstract The technical developments in recent years have brought the quantity and quality of digital information to an unprecedented level, as enormous archives of satellite images are available to the users. However, even if these advances open more and more possibilities in the use of digital imagery, they also rise several problems of storage and treatment. The latter is considered in this Thesis: the processing of very high spatial and spectral resolution images is treated with approaches based on data-driven algorithms relying on kernel methods. In particular, the problem of image classification, i.e. the categorization of the image's pixels into a reduced number of classes reflecting spectral and contextual properties, is studied through the different models presented. The accent is put on algorithmic efficiency and the simplicity of the approaches proposed, to avoid too complex models that would not be used by users. The major challenge of the Thesis is to remain close to concrete remote sensing problems, without losing the methodological interest from the machine learning viewpoint: in this sense, this work aims at building a bridge between the machine learning and remote sensing communities and all the models proposed have been developed keeping in mind the need for such a synergy. Four models are proposed: first, an adaptive model learning the relevant image features has been proposed to solve the problem of high dimensionality and collinearity of the image features. This model provides automatically an accurate classifier and a ranking of the relevance of the single features. The scarcity and unreliability of labeled. information were the common root of the second and third models proposed: when confronted to such problems, the user can either construct the labeled set iteratively by direct interaction with the machine or use the unlabeled data to increase robustness and quality of the description of data. Both solutions have been explored resulting into two methodological contributions, based respectively on active learning and semisupervised learning. Finally, the more theoretical issue of structured outputs has been considered in the last model, which, by integrating outputs similarity into a model, opens new challenges and opportunities for remote sensing image processing.

A case study of landslides and coping strategies in two villages of central-eastern Nepal

Landslides are an increasing problem in Nepal's Middle Hills due to both natural and human phenomena: mainly increasingly intense monsoon rains and a boom in rural road construction. This problem has largely been neglected due to underreporting of losses and the dispersed nature of landslides. Understanding how populations cope with landslides is a first step toward developing more effective landslide risk management programs. The present research focuses on two villages in Central-Eastern Nepal, both affected by active landslides but with different coping strategies. Research methods are interdisciplinary, based on a geological assessment of landslide risk and a socio-economic study of the villages using household questionnaires, focus group discussions and transect walks. Community risk maps are compared with geological landslide risk maps to better understand and communicate community risk perceptions, priorities and coping strategies. A modified typology of coping strategies is presented, based on previous work by Burton, Kates, and White (1993) that is useful for decision-makers for designing more effective programs for landslide mitigation. Main findings underscore that coping strategies, mainly seeking external assistance and outmigration, are closely linked to access to resources, ethnicity/social status and levels of community organization. Conclusions include the importance of investing in organizational skills, while building on local knowledge about landslide mitigation for reducing landslide risk. There is great potential to increase coping strategies by incorporating skills training on landslide mitigation in existing agricultural outreach and community forest user group training.

Learning about Bayesian networks for forensic interpretation : An example based on the "the problem of multiple propositions"

Both, Bayesian networks and probabilistic evaluation are gaining more and more widespread use within many professional branches, including forensic science. Notwithstanding, they constitute subtle topics with definitional details that require careful study. While many sophisticated developments of probabilistic approaches to evaluation of forensic findings may readily be found in published literature, there remains a gap with respect to writings that focus on foundational aspects and on how these may be acquired by interested scientists new to these topics. This paper takes this as a starting point to report on the learning about Bayesian networks for likelihood ratio based, probabilistic inference procedures in a class of master students in forensic science. The presentation uses an example that relies on a casework scenario drawn from published literature, involving a questioned signature. A complicating aspect of that case study - proposed to students in a teaching scenario - is due to the need of considering multiple competing propositions, which is an outset that may not readily be approached within a likelihood ratio based framework without drawing attention to some additional technical details. Using generic Bayesian networks fragments from existing literature on the topic, course participants were able to track the probabilistic underpinnings of the proposed scenario correctly both in terms of likelihood ratios and of posterior probabilities. In addition, further study of the example by students allowed them to derive an alternative Bayesian network structure with a computational output that is equivalent to existing probabilistic solutions. This practical experience underlines the potential of Bayesian networks to support and clarify foundational principles of probabilistic procedures for forensic evaluation.

Bayesian Networks and the Value of the Evidence for the Forensic Two-Trace Transfer Problem

Forensic scientists face increasingly complex inference problems for evaluating likelihood ratios (LRs) for an appropriate pair of propositions. Up to now, scientists and statisticians have derived LR formulae using an algebraic approach. However, this approach reaches its limits when addressing cases with an increasing number of variables and dependence relationships between these variables. In this study, we suggest using a graphical approach, based on the construction of Bayesian networks (BNs). We first construct a BN that captures the problem, and then deduce the expression for calculating the LR from this model to compare it with existing LR formulae. We illustrate this idea by applying it to the evaluation of an activity level LR in the context of the two-trace transfer problem. Our approach allows us to relax assumptions made in previous LR developments, produce a new LR formula for the two-trace transfer problem and generalize this scenario to n traces.

La collaboration interprofessionnelle à l'interface des institutions socio-éducative et psychiatrique. Jeux et enjeux de l'intervention auprès de personnes en situation de handicap mental et de troubles psychiques

La prise en charge et le suivi de personnes en situation de handicap mental souffrant de troubles psychiques et se trouvant donc à l'interface des domaines socio:éducatif et psychiatrique, constituent des défis complexes en matière de collaboration interprofessionnelle. Dans le canton de Vaud, les acteurs concernés par ce problème s'efforcent depuis de nombreuses années de créer des réseaux pluridisciplinaires visant un meilleur échange entre professionnels et le développement de compétences et de connaissances permettant d'améliorer le bien:être des bénéficiaires. Ce travail se propose ainsi d'étudier et de questionner ces modalités de travail dans une perspective socioculturelle (Vygotski, 1934/1997), afin d'en comprendre le fonctionnement, d'en éclairer les mécanismes et de fournir des pistes de réflexion aux professionnels. Il repose sur un travail de terrain mené auprès des membres du Dispositif de Collaboration Psychiatrie Handicap Mental (DCPHM) du Département de psychiatrie du CHUV, dont la mission principale est de faciliter la collaboration entre les institutions socio:éducatives et psychiatriques spécialisées dans le suivi des personnes en situation de handicap mental et souffrant de troubles psychiques. Le travail empirique est basé sur une approche qualitative et compréhensive des interactions sociales, et procède par une étude de terrain approfondie. Les données recueillies sont variées : notes de terrain et récolte de documentation, enregistrement de réunions d'équipe au sein du DCPHM et de réunions de réseau, et entretiens de différents types. L'analyse montre que le travail de collaboration qui incombe à l'équipe est constitué d'obstacles qui sont autant d'occasions de développement professionnel et de construction identitaire. Les résultats mettent en lumière des mécanismes discursifs de catégorisation concourant à la fois à la construction des patients comme objets d'activité, et à la construction d'une place qui légitime les interventions de l'équipe dans le paysage socio:éducatif et psychiatrique vaudois et la met au centre de l'arène professionnelle. -- Care and follow:up for people with mental disabilities suffering from psychological disorders : therefore at the interface between the socio:educational and psychiatric fields : represent complex challenges in terms of interprofessional collaboration. In the canton of Vaud, the caregivers involved in this issue have been trying for years to build multidisciplinary networks in order to better exchange between professionals and develop skills and knowledge to improve the recipients' well:being. This work thus proposes to study and question these working methods in a sociocultural perspective (Vygotski, 1934/1997) so as to understand how they operate, highlight inherent mechanisms and provide actionable insights to the professionals. It is based on fieldwork conducted among members of the Dispositif de Collaboration Psychiatrie Handicap Mental (DCPHM), of the Psychiatry Department at the CHUV University Hospital in Lausanne, whose main mission is to facilitate collaboration between the socio:educational and psychiatric institutions specialising in monitoring people presenting with both mental handicap and psychiatric disorder. The empirical work is based on a qualitative and comprehensive approach to social interactions, and conducted based on an in:depth field study. The data collected are varied - field notes and documentation collection, recordings of team meetings within the DCPHM and network meetings, and various types of interviews. The analysis shows that the collaborative work that befalls the team consists of obstacles, all of which provide opportunities for professional development and identity construction. The results highlight discursive strategies of categorisation which contribute both to the construction of the patients as objects of activity and to building a position that legitimates the team's interventions in the socio: educational and psychiatric landscape of canton Vaud and puts it in the centre of the professional arena.