Machine learning and data decompositions for complex networked dynamical systems

Thesis (Ph.D.)--University of Washington, 2016-08

Essays on Machine Learning and Hedonic Models

Thesis (Ph.D.)--University of Washington, 2016-08

Adapting to a team of host culture domination: Experiences of Finnish and British expatriates

International teams are permanent features of the global market, and multinational companies cannot ignore them. This thesis discusses an individual’s adaptation to a team of host culture domination by looking at the experiences of Finnish and British expatriates. The research question is: How does a foreigner adapt to a Finnish-British team where all the other members are from the local culture? The theoretical framework of this thesis consists of theories on multicultural teamwork and Finnish and British cultures. Based on theories, the impact of diversity on teamwork and whether it improves or deteriorates a team’s performance is unclear but the idea of multicultural teams is not just to live with cultural differences but to use and benefit of them. The main differences between Finnish and British cultures are in management styles and level of hierarchy. Still, these two cultures seem to be very similar. This is a qualitative study and expert interviews were used as a method for data collection. The results of the empirical research showed that Finns and Brits have differences in decision-making, level of hierarchy and communication. None of the informants saw the challenges that they have faced to result from being the only representative of a foreign culture – they only perceived them resulting from the characteristics of the host culture. In addition, all the informants had to change their habits and behaviour in the host culture in order to fit into the new culture and to become a part of the team. However, all the informants saw advantages in cross-cultural teams. Based on this thesis, team members from the host culture do not value cultural differences and they seem to be an issue only for the team member from the minority culture. It seems that a Finn or a Brit adapts to a Finnish-British team where the host culture dominates by changing their own behaviour in order to become a part of the team – even though people often see their own culture’s methods and habits as the right ones. Finnish and British cultures seem to be similar in fundamental issues like views and attitudes, which could mean that it is fairly easy for a Finn to get used to working in a British team and vice versa. Even though there are many challenges in cross-cultural and bi-cultural teams, there are even more advantages.

Political Strategy, Leadership, and Policy Entrepreneurship in Japanese Defense Policy and Politics: A Comparison of Three Prime Ministerships

Since the end of the Cold War, Japan’s defense policy and politics has gone through significant changes. Throughout the post cold war period, US-Japan alliance managers, politicians with differing visions and preferences, scholars, think tanks, and the actions of foreign governments have all played significant roles in influencing these changes. Along with these actors, the Japanese prime minister has played an important, if sometimes subtle, role in the realm of defense policy and politics. Japanese prime ministers, though significantly weaker than many heads of state, nevertheless play an important role in policy by empowering different actors (bureaucratic actors, independent commissions, or civil actors), through personal diplomacy, through agenda-setting, and through symbolic acts of state. The power of the prime minister to influence policy processes, however, has frequently varied by prime minister. My dissertation investigates how different political strategies and entrepreneurial insights by the prime minister have influenced defense policy and politics since the end of the Cold War. In addition, it seeks to explain how the quality of political strategy and entrepreneurial insight employed by different prime ministers was important in the success of different approaches to defense. My dissertation employs a comparative case study approach to examine how different prime ministerial strategies have mattered in the realm of Japanese defense policy and politics. Three prime ministers have been chosen: Prime Minister Hashimoto Ryutaro (1996-1998); Prime Minister Koizumi Junichiro (2001-2006); and Prime Minister Hatoyama Yukio (2009-2010). These prime ministers have been chosen to provide maximum contrast on issues of policy preference, cabinet management, choice of partners, and overall strategy. As my dissertation finds, the quality of political strategy has been an important aspect of Japan’s defense transformation. Successful strategies have frequently used the knowledge and accumulated personal networks of bureaucrats, supplemented bureaucratic initiatives with top-down personal diplomacy, and used a revitalized US-Japan strategic relationship as a political resource for a stronger prime ministership. Though alternative approaches, such as those that have looked to displace the influence of bureaucrats and the US in defense policy, have been less successful, this dissertation also finds theoretical evidence that alternatives may exist.

Methods for integrating machine learning and constrained optimization

In the framework of industrial problems, the application of Constrained Optimization is known to have overall very good modeling capability and performance and stands as one of the most powerful, explored, and exploited tool to address prescriptive tasks. The number of applications is huge, ranging from logistics to transportation, packing, production, telecommunication, scheduling, and much more. The main reason behind this success is to be found in the remarkable effort put in the last decades by the OR community to develop realistic models and devise exact or approximate methods to solve the largest variety of constrained or combinatorial optimization problems, together with the spread of computational power and easily accessible OR software and resources. On the other hand, the technological advancements lead to a data wealth never seen before and increasingly push towards methods able to extract useful knowledge from them; among the data-driven methods, Machine Learning techniques appear to be one of the most promising, thanks to its successes in domains like Image Recognition, Natural Language Processes and playing games, but also the amount of research involved. The purpose of the present research is to study how Machine Learning and Constrained Optimization can be used together to achieve systems able to leverage the strengths of both methods: this would open the way to exploiting decades of research on resolution techniques for COPs and constructing models able to adapt and learn from available data. In the first part of this work, we survey the existing techniques and classify them according to the type, method, or scope of the integration; subsequently, we introduce a novel and general algorithm devised to inject knowledge into learning models through constraints, Moving Target. In the last part of the thesis, two applications stemming from real-world projects and done in collaboration with Optit will be presented.

Machine learning and convolutional neural networks for melanomas identification

Il machine learning negli ultimi anni ha acquisito una crescente popolarità nell’ambito della ricerca scientifica e delle sue applicazioni. Lo scopo di questa tesi è stato quello di studiare il machine learning nei suoi aspetti generali e applicarlo a problemi di computer vision. La tesi ha affrontato le difficoltà del dover spiegare dal punto di vista teorico gli algoritmi alla base delle reti neurali convoluzionali e ha successivamente trattato due problemi concreti di riconoscimento immagini: il dataset MNIST (immagini di cifre scritte a mano) e un dataset che sarà chiamato ”MELANOMA dataset” (immagini di melanomi e nevi sani). Utilizzando le tecniche spiegate nella sezione teorica si sono riusciti ad ottenere risultati soddifacenti per entrambi i dataset ottenendo una precisione del 98% per il MNIST e del 76.8% per il MELANOMA dataset

Of housing and politics : mapping political opportunities for mobilising in Bangalore, India

Quelles sont les conditions pour l'émergence d'une mobilisation sociale en faveur du logement convenable dans la métropole de Bangalore (Inde)? Cette question, qui est au coeur de cette thèse, est particulièrement pertinente dans le contexte d'une ville où 1,7 million de personnes, soit un cinquième de la population, vit dans des bidonvilles. L'absence d'un mouvement mettant en cause l'échec des politiques publiques du logement est intéressante dans la mesure où l'Inde a hérité un système de gouvernance colonial et d'une tradition de mouvements sociaux. Pour répondre à ce questionnement, un cadre théorique issu de la littérature sur les mouvements sociaux est développé. Il s'articule autour des liens entre les opportunités politiques au niveau macro et les répertoires d'action des organisations de mouvement social (OMS) au niveau méso, de la tension entre la formalité de la loi et des politiques publiques et l'informalité des circuits d'échange, de la corruption et du clientélisme, et enfin, se focalise sur les systèmes de discours de caste et de la citoyenneté et de leur concrétisation dans des systèmes d'organisations et de réseaux sociaux. Ce cadre théorique permet d'étudier empiriquement la question à travers quatre OMS dans la ville de Bangalore. Les résultats mettent en avant l'existence de mécanismes complexes. Les opportunités politiques formelles n'étant ouvertes que sur le plan rhétorique, elles ne peuvent être véritablement utilisées que par des moyens légaux ou contentieux, ce qui nécessite des compétences sociales dont la plupart des habitants des bidonvilles sont dépourvus. L'inadéquation entre les ressources à disposition pour les logements sociaux et les besoins très importants des pauvres, donne un poids politique considérable aux acteurs en charge de l'attribution de ces ressources rares. Cet état de fait a des répercussions sur la politique électorale. Les habitants des bidonvilles représentant un poids électoral important, ils sont mobilisés à travers de pratiques clientélistes. La corruption et le clientélisme se nourrissent mutuellement pour maintenir une certaine dépendance des habitants. Les OMS qui développent un répertoire discursif remettant en cause le système de caste et qui encouragent une conscience citoyenne, se sont avérées les plus durables pour résister à la cooptation des forces politiques. Cette recherche empirique met en lumière l'inadéquation entre les prescriptions formelles dans le domaine de la gouvernance des besoins humains, tels que le logement, et les pratiques réelles sur le terrain. Cette recherche appelle à réfléchir au-delà de la diffusion du discours sur la « bonne gouvernance » vers des formes de « gouvernance vernaculaire » qui prendrait au sérieux l'informalité en développant une compréhension des avantages à court terme pour les personnes marginalisées dans la ville et les effets à long terme sur la pratique démocratique. - What are the conditions for the emergence of a social movement on the issue of adequate housing in the metropolitan city of Bangalore (India)? This question is at the heart of this dissertation and is particularly pertinent against the background that an estimated 1.7 million or about 20% of the city's population lives in slums. The absence of a movement addressing the failure of public housing policy despite India having inherited colonial systems of governance and traditions of movement is noteworthy. Answers are sought within a theoretical framework stemming from social movement theories that incorporates three linkages articulating around: Macro-level political opportunities and meso-level action repertoires of social movement organisations (SMOs), tensions between the formality of law, policy and the informality of exchange circuits of corruption and clientelism and finally around systems of discourses of caste and citizenship and their instantiation in concrete systems of social organisations and networks. This thesis is empirically investigated through a qualitative case study research design involving four sampled social movement organisations. The results bring complex mechanisms to the fore. Formal political opportunities are only rhetorically open and have to be cracked through legal weaponry or contentious escalation, which requires considerable social skills that slum-dwellers often lack. The inadequacy between the few housing resources and the vast number of slum-dwellers transform housing benefits and urban service provisions into political currency. Such a state of affairs has serious repercussions on conditions for mobilisation. They become imbricated with electoral logic, in which slum-dwellers represent large vote-banks and where corruption and clientelism feed each other to maintain a certain dependency of the poor. SMOs deploying a discursive repertoire that questioned the caste system and encouraged a pursuit of citizenship proved to be the most sustainable to resist co-option from political forces. This empirical investigation brings to light the mismatch between the formal prescriptions in the domain of the governance of basic human needs such as housing and the real practices on the ground. This research calls to reflect beyond the inadequacy of the diffused « good governance » discourse towards forms of « vernacular governance » that take informality seriously in understanding the short-term benefits for the marginalised in the city and the long-term effects on democratic practice.

Workplace aggression in organizational changing processes: The mediation role of leader political (soft and hard) behavior

One of the main trends in workplace aggression research is studying its antecedents. But the literature also reveals that some predictors remain understudied, like organizational change [1]. Additionally, possible mediators of this relationship were not investigated. The main objective of this research is studding the mediating effect of the leader political behavior (soft and hard version) on the relationship between organizational change and workplace aggression. Participants representing a wide variety of jobs across many organizations were surveyed. The measures used in this research are an Organizational Change Questionnaire climate of change, processes, and readiness [2], a Workplace Aggression Scale [e.g. 3, 4] and a Political Behavior Questionnaire [5]. The results of the study and its theoretical and practical implications will be presented and discussed.

Kitchen robot prod. line development. Proj. of Bowden cables cutting and thermal forming machine.

This document presents particular description of work done during student’s internship in PR Metal company realized as ERASMUS PROJECT at ISEP. All information including company’s description and its structure, overview of the problems and analyzed cases, all stages of projects from concept to conclusion can be found here. Description of work done during the internship is divided here into two pieces. First part concerns one activities of the company which is robotic chefs (kitchen robot) production line. Work, that was done for development of this line involved several tasks, among them: creating a single-worker montage station for screwing robots housing’s parts, improve security system for laser welding chamber, what particularly consists in designing automatically closing door system with special surface, that protects against destructive action of laser beam, test station for examination of durability of heating connectors, solving problem with rotors vibrations. Second part tells about main task, realized in second half of internship and stands a complete description of machine development and design. The machine is a part of car handle latch cable production line and its tasks are: cutting cable to required length and hot-forming plastic cover for further assembly needs.

Discrimination of Brazilian propolis according to the seasoning using chemometrics and machine learning based on UV-Vis scanning data

Propolis is a chemically complex biomass produced by honeybees (Apis mellifera) from plant resins added of salivary enzymes, beeswax, and pollen. The biological activities described for propolis were also identified for donor plants resin, but a big challenge for the standardization of the chemical composition and biological effects of propolis remains on a better understanding of the influence of seasonality on the chemical constituents of that raw material. Since propolis quality depends, among other variables, on the local flora which is strongly influenced by (a)biotic factors over the seasons, to unravel the harvest season effect on the propolis chemical profile is an issue of recognized importance. For that, fast, cheap, and robust analytical techniques seem to be the best choice for large scale quality control processes in the most demanding markets, e.g., human health applications. For that, UV-Visible (UV-Vis) scanning spectrophotometry of hydroalcoholic extracts (HE) of seventy-three propolis samples, collected over the seasons in 2014 (summer, spring, autumn, and winter) and 2015 (summer and autumn) in Southern Brazil was adopted. Further machine learning and chemometrics techniques were applied to the UV-Vis dataset aiming to gain insights as to the seasonality effect on the claimed chemical heterogeneity of propolis samples determined by changes in the flora of the geographic region under study. Descriptive and classification models were built following a chemometric approach, i.e. principal component analysis (PCA) and hierarchical clustering analysis (HCA) supported by scripts written in the R language. The UV-Vis profiles associated with chemometric analysis allowed identifying a typical pattern in propolis samples collected in the summer. Importantly, the discrimination based on PCA could be improved by using the dataset of the fingerprint region of phenolic compounds ( = 280-400m), suggesting that besides the biological activities of those secondary metabolites, they also play a relevant role for the discrimination and classification of that complex matrix through bioinformatics tools. Finally, a series of machine learning approaches, e.g., partial least square-discriminant analysis (PLS-DA), k-Nearest Neighbors (kNN), and Decision Trees showed to be complementary to PCA and HCA, allowing to obtain relevant information as to the sample discrimination.

Metabolic profiling and classification of propolis samples from Southern Brazil: an NMR-based platform coupled with machine learning

The chemical composition of propolis is affected by environmental factors and harvest season, making it difficult to standardize its extracts for medicinal usage. By detecting a typical chemical profile associated with propolis from a specific production region or season, certain types of propolis may be used to obtain a specific pharmacological activity. In this study, propolis from three agroecological regions (plain, plateau, and highlands) from southern Brazil, collected over the four seasons of 2010, were investigated through a novel NMR-based metabolomics data analysis workflow. Chemometrics and machine learning algorithms (PLS-DA and RF), including methods to estimate variable importance in classification, were used in this study. The machine learning and feature selection methods permitted construction of models for propolis sample classification with high accuracy (>75%, reaching 90% in the best case), better discriminating samples regarding their collection seasons comparatively to the harvest regions. PLS-DA and RF allowed the identification of biomarkers for sample discrimination, expanding the set of discriminating features and adding relevant information for the identification of the class-determining metabolites. The NMR-based metabolomics analytical platform, coupled to bioinformatic tools, allowed characterization and classification of Brazilian propolis samples regarding the metabolite signature of important compounds, i.e., chemical fingerprint, harvest seasons, and production regions.

Do political incentives matter for tax policies? Ideology, opportunism and the tax structure

This paper investigates the importance of political ideology and opportunism in the choice of the tax structure. In particular, we examine the effects of cabinet ideology and elections on the distribution of the tax burden across factors of production and consumption for 21 OECD countries over the period 1970-2000 by employing four alternative cabinet ideology measures and by using the methodology of effective tax rates. There is evidence of both opportunistic and partisan effects on tax policies. More precisely, we find that left-wing governments rely more on capital relative to labor income taxation and that they tend to increase consumption taxes. Moreover, we find that income tax rates (but not consumption taxes) tend to be reduced in preelectoral periods and that capital effective tax rates (defined broadly to include taxes on selfemployed income) are reduced by more than effective labor tax rates.