97 research outputs found

    IBPP Research Associates: Namibia/Turkey

    No full text
    This article - The Globalization of Old Age, by Serdan Sayan - was posted on the January 9, 2003 issue of The Namibian (Namibia). It has not been posted in Scholarly Commons due to copyright restrictions. Though a subscription for full access to the article is required, partial access to the article is available online in the archives of Project Syndicate. The article discusses wealth inequality and the aging, or graying, of the world\u27s richest nations and the subsequent possible global impact of this demographic reality. The author is Vice-Chairman and Director of Graduate Studies, Department of Economics, Bilkent University, Ankara, Turkey

    Vision équivariante en mode non supervisée et en présence d'une faible quantité de données

    No full text
    This thesis addresses the key issue of creating robust vision models that learn efficientrepresentations in unsupervised, low-data regimes and are geometrically consistent underspatial transformations. Conventional convolutional neural networks only have a nascentaugmentation, with large parameter spaces to ensure robustness in transformations like rotation and scaling. Even then they are highly complicit to aliasing issues, group convolutions however provide a more principled way to preserve equivariance. However, they are subject to severe computational impediments in autoencoding settings for downstream tasks, owing to dimensional expansions caused by lifting operations, memory-based limitations, as well as pooling operations that can often break equivariance. In our work we explore the frequency domain, by presentinga common theoretical framework for autoencoding. This extends to equivariant objectdetection and anomaly localization formally connecting, group equivariance with Rieszrepresentation theory. We establish its superior generalization abilities with tighter PAC-Bayesian bounds. Our novel LeaRN-EqSTN architecture performs sequential estimationof transformations via a learnable Riesz transform and spatial transformer networks withthe aim of learning equivariance. This approach achieves greater computational efficiencywithout sacrificing theoretical guarantees. This architecture allows direct integration to autoencoding models while preserving equivariance properties. The thesis presents a number of main contributions. We present SPAGMACE, a novel unsupervised glimpse-based object detection system that organizes the latent space with Gaussian mixture priors to foster improved semantic interpretation. We extend LeaRN-EqSTN with SPAGMACE to develop the first glimpse-based equivariant unsupervised object detection model with enhanced performance on real-world datasets. To facilitate anomaly localization, we suggest an autoencoder framework that combines the strength of our LeaRN-EqSTN model with an efficient post-processing able to discriminate anomalies from normal non-rigid distortions. We demonstrate the effectiveness our approach in all three setting: autoencoding, unsupervised object detection and anomaly localization. Our architectures show superior generalization compared to state-of-the-art models, particularly in low-data regimes. This work demonstrates that incorporating geometric equivariance based on frequency level representations into neural architectures provides a principled approach to learning robust visual representations from limited data. Theoretical study unveils the inherent relationships among symmetry, aliasing, and generalization, while empirical results demonstrate improvements over state-of-the-art techniques in computer vision.Cette thèse aborde la question clé de la création de modèles de vision robustes qui apprennent des représentations efficaces dans des régimes non supervisés à faible quantité de données et qui sont géométriquement cohérents sous des transformations géométriques rigides. Les réseaux neuronaux à convolution classiques ne possèdent qu'une augmentation naissante, avec de grands espaces de paramètres pour garantir la robustesse face à des transformations telles que la rotation et la mise à l'échelle. Même dans ce cas, ils sont fortement sujets aux problèmes de repliement spectral. Les convolutions de groupe, en revanche, offrent une manière plus rigoureuse de préserver l'équivariance. Cependant, elles sont confrontées à de sérieux obstacles computationnels dans des contextes d'autoencodage pour des tâches en aval, en raison des expansions dimensionnelles causées par les opérations de lifting, des limitations basées sur la mémoire, ainsi que des opérations de pooling qui peuvent souvent briser l'équivariance. Dans notre travail, nous explorons le domaine fréquentiel en présentant un cadre théorique commun pour l'autoencodage. Cela s'étend à la détection équivariante d'objets et à la localisation d'anomalies, reliant formellement l'équivariance de groupe à la théorie de la représentation de Riesz. Nous établissons ses capacités de généralisation supérieures avec des bornes PAC-bayésiennes plus resserrées. Notre nouvelle architecture LeaRN-EqSTN effectue une estimation séquentielle des transformations via une transformée de Riesz apprenable et des réseaux de transformation spatiale dans le but d'apprendre l'équivariance. Cette approche atteint une plus grande efficacité computationnelle sans sacrifier les garanties théoriques. Cette architecture permet une intégration directe aux modèles d'autoencodage tout en préservant les propriétés d'équivariance. La thèse présente plusieurs contributions principales. Nous proposons SPAGMACE, un nouveau système de détection d'objets basé sur des aperçus non supervisés qui organise l'espace latent avec des a priori de mélange gaussien pour favoriser une meilleure interprétation sémantique. Nous étendons LeaRN-EqSTN avec SPAGMACE pour développer le premier modèle de détection d'objets équivariant non supervisé basé sur des aperçus, avec des performances améliorées sur des ensembles de données réelles. Pour faciliter la localisation d'anomalies, nous suggérons un cadre d'autoencodeur qui combine la force de notre modèle LeaRN-EqSTN avec un post-traitement efficace capable de discriminer les anomalies des distorsions non rigides normales. Nous démontrons l'efficacité de notre approche dans trois contextes : l'autoencodage ainsi que la détection d'objets et la localisation d'anomalies non supervisées. Nos architectures montrent une généralisation supérieure par rapport aux modèles de l'état de l'art, en particulier dans les situations où la quantité de données est faible. Ce travail démontre que l'incorporation de l'équivariance géométrique basée sur des représentations au niveau fréquentiel dans les architectures neuronales offre une approche rigoureuse pour apprendre des représentations visuelles robustes à partir de données en quantité limitée. Une étude théorique révèle les relations inhérentes entre la symétrie, le repliement spectral et la généralisation, tandis que les résultats empiriques démontrent des améliorations par rapport aux techniques de de l'état de l'art en vision par ordinateur

    PENERAPAN KONSEP COMFORT OF HUMAN CIRCULATION DAN ARSITEKTUR NEO VERNAKULAR PADA REDESAIN PASAR TRADISIONAL DESA SAYAN

    No full text
    Traditional markets are places where buying and selling transactions take place between traders and buyers by implementing a bargaining system that is characteristic of traditional markets. Many people depend on the existence of traditional markets. Especially for lower middle class people. Sayan Village Traditional Market, Ubud District, Gianyar Regency is a location for trading daily necessities such as fruit, vegetables, meat, prayer tools, and others. The problems faced by this market are unhygienic market conditions and the appearance of slums, buildings and spatial layout. Sayan Market experiences an inadequate and disorganized condition, with a number of areas not being utilized optimally, such as chaotic parking and kiosk facilities and main market stalls that are neglected and not well maintained. This can cause people to feel uncomfortable, and people prefer modern markets that prioritize cleanliness. The research method used by the author to prepare this research is a qualitative method in the form of literature study, observation, documentation and interviews. Based on the existing problems, it is necessary to redesign the Traditional Market in Sayan Village, Ubud District, Gianyar Regency so that it can accommodate functions well packaged in a new, clean, comfortable market appearance and pay attention to more optimal circulatio

    Moss occurrences in Salair-Kuznetsk Region (Altai-Sayan mountain country)

    No full text
    In the flora of large regions, mosses comprise about a quarter of the total diversity of higher plants. However, now mosses are the least studied group of higher plants. Data on moss species distribution are fragmentary, especially in Russia with its vast expanse and low density of botanists.The author for many years has been studying the bryoflora of various areas of the Salair-Kuznetsk Region. In addition to the herbarium collection, the author's bryological relevés were organised as a Database. It stores all the assembled information for the years about the locations of the species, including when the specimens were not placed in the herbarium.The article describes three datasets that were arranged from the author's databases for three geomorphological units in the northwest of Altai-Sayan mountain country (South Siberia). Together, these three units are combined into the Salair-Kuznetsk Region. The datasets are:1. Moss occurrences in the Kuznetsk upland.The dataset consists of 3940 occurrence records and includes both preserved specimens (1135) and ‘human observations’ of the author (2805). The material was collected mainly from 1992-2011; some samples collected by A. N. Vasiliev in 1970-1971 (165) were also taken into account. A total of 312 moss taxa belonging to 135 genera and 41 families are reported for the region.2. Moss occurrences in Salair Ridge.The dataset consists of 2442 occurrence records and includes both preserved specimens (553) and ‘human observations’ of the author (1889). The material was collected mainly from 1992-1996; a total of 231 moss taxa belonging to 119 genera and 35 families are reported for the region.3. Moss occurrences in Kuznetsk Depression.The dataset consists of 1690 occurrence records and includes both preserved specimens (281) and ‘human observations’ of the author (1409). The material was collected mainly from 2007-2014; a total of 155 moss taxa belonging to 85 genera and 30 families are reported for the region.All the records are geo-linked. The uncertainty of coordinates in metres varies from 500.0-10000.0 m for the earliest records that are geo-linked by topo-map, to 10.0-100.0 m for records after 2003 that are geo-linked by GPS.The article summarises the results of the author's long-term bryological investigations in the Salair-Kuznetsk Region (northwest of Altai-Sayan mountain country, South Siberia).In total, 8072 occurrence records for 366 moss species from 148 genera and 41 families are published for the territory. The datasets contribute to filling gaps in the moss species distribution and ecology

    Epiphenomenalism and Human Sexuality

    No full text
    The paper is a cognitive and statistical study of sexuality based on the grounds of Epiphenomalism, the paper also proposes some important ideas and proves some important results, statistically based on two surveys

    POSITION OF PRADANA IN NYEBURIN MARRIAGE REVIEWED FROM THE LAW OF BALI INDIGENOUS HERITAGE IN BANJAR KUTUH DESA SAYAN KECAMATAN UBUD KABUPATEN GIANYAR

    No full text
    The aims of this study are to know the Nyeburin Marriage Inheritance System in Banjar Kutuh, Sayan Village, Ubud District, Gianyar Regency women as Purusa or nyentana men as Pradana if their inheritance is in the form of land assets and to know the position of social social status of men as Pradana in Nyeburin Marriage in indigenous communities in the Village of Sayan Banjar Kutuh, Ubud District, Gianyar Regency. This method uses Empirical legal research. Using the statutory approach, case approach, and conceptual approach. Sources of data are sourced from primary data (field research) and secondary data (library research). Data collection techniques consisted of interview techniques, document study techniques, and literature study techniques. The data collected was analyzed descriptive qualitatively. The author uses the theory of justice, the theory of legal certainty, the theory of reception in complex u. Based on the results of the study it was found that the inheritance of nyeburin marriage in Banjar village is still strong in women because of their status as Purusa and their social social status is different if the work of a person who becomes a pradana, for example, a doctor may look higher. The conclusion is that a woman still has the right to inherit because as a purusa and if the inheritance in the form of land remains the right of the woman and in social status remains the head of the household in the community remains the same except for work that makes their social status different

    Special Issue on “Recent Advances in Robust Adaptive Control”

    No full text
    Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Team Bart De Schutte

    The HathiTrust Digital Library’s potential for musicology research

    No full text
    The HathiTrust Digital Library (HTDL) is one of the largest digital libraries in the world, containing over fourteen million volumes from the collections of major academic and research libraries. In this paper, we discuss the HTDL's potential for musicology research by providing a bibliometric analysis of the collection as a whole, and of the music materials in particular. A series of case studies illustrates the kinds of musicological research that may be conducted using the HTDL. We highlight several opportunities for improvement, and discuss promising future directions for new knowledge creation through the processing and analysis of large amounts of retrospective data. The HTDL presents significant new opportunities to the study of music that will continue to expand as data, metadata and collection enhancements are introduced

    THE ISSUES OF THE USE OF INDIGENOUS SACRAL OBJECTS IN TOURIST INDUSTRY (EXPERIENCE OF THE ALTAI-SAYAN REGION)

    No full text
    &lt;p&gt;Religious objects play an important role in the vital activity of the indigenous peoples of the North, as they accumulate information about the history, traditional culture and the use of local communities. Currently, many of them are gradually fading away or in an unsatisfactory state. Sacred places are getting more and more popular with tourists, which turns them into touristic destinations. That is why the article raises the pressing challenge of including the sacred places of the Northern indigenous peoples into tourist routes.&lt;strong&gt;&lt;/strong&gt;&lt;/p&gt;&lt;p&gt;The article examines the experience of the regions ofSouthern Siberiain organizing tourism near religious sites.&lt;/p&gt;&lt;p&gt;As a result of the research, the author comes to the conclusion that the development of tourism in places of power of the indigenous peoples of the North has both positive and negative aspects. The paper gives a number of practical recommendations that would prevent negative consequences of the use of religious objects in tourism. In the conclusion of the article it is said that, with proper organization, tourism can be one of the most effective tools for maintaining and developing the sacred sites in the Altai-Sayan region.&lt;/p&gt;</jats:p
    corecore