5916 research outputs found
Sort by
Statistically improving K-means clustering performance
No full textThe traditional K-means algorithm is a cornerstone in unsupervised learning, providing a simple yet effective method for data clustering. However, its reliance on random initialization often leads to sub-optimal clustering results. This paper introduces an enhanced version of K-means algorithm, aimed at improving the clustering results. Our proposed methodology depends on identifying clusters that need to be partitioned and clusters that need to be merged through a series of statistical operation and iteratively resolve the problem leading to better clusters. We compare our proposed approach to K-Means and K-means++ algorithms on S-2, California housing prices, and EMNIST datasets showing performance improvements
Biosensing with NV-centers in optically trapped nanodiamonds
No full textNanodiamonds with color centers are to an increasing degree investigated as intracellular biosensors for magnetic fields, electrical fields, or temperature changes, as well as for abundance of free radicals or pH inside live cells. A common color-center is the nitrogen-vacancy (NV) center, where a substitutional nitrogen atom is positioned next to a vacancy within the diamond host crystal. The nanodiamond with negatively charged NV- center is particularly versatile due to its biocompatibility and its purely optical addressability. Our work aims to use NV-center nanodiamonds both as intracellular biosensors and as probe particles within an optical trap to determine the viscoelastic properties of the intracellular environment in single-cell studies. For this to be successful, several prior steps are needed: 1) The uptake of nanodiamonds within the cells should be characterized, including studies of subcellular localization, and a controllable protocol developed; 2) Any effect of the trapping laser on the NV-center sensing should be characterized and understood, and a protocol for stable trapping along with accurate biosensing should be developed. In this work we summarize the preliminary findings of our ongoing investigations to address these points. We show results of T1-relaxometry with and without CW NIR laser irradiation in a suspension cell model, analyze optical trapping of nanodiamonds with CW light in an adhesion cell model, and investigate implications of the presence of the optical trapping laser on T1-relaxometry measurements.Det Frie Forskningsrad (DFF) ; Novo Nordisk Foundation ; H2020 Marie Skłodowska-Curie Action
Experimental and numerical analysis of frost formation over a flat additive manufactured surface
No full textFrost formation is a major challenge in the design of heat exchanger components in refrigeration and heat pump systems. The frost layer on the surfaces acts as an insulation layer on the surface which blocks the heat transfer to the surface and reduces the heat transfer capacity of the heat exchanger. Additive manufacturing (AM) has been of interest over the last decade for a wide range of engineering applications and offers certain advantages in the manufacturing of complex geometries.
In this study, a custom-designed AM surface with pins is manufactured and experimentally tested in a closed-loop frosting wind tunnel. The effect of pinned fin geometry of AM plate on frost accumulation is examined. According to the comparison, the influence of the water vapor diffusivity coefficient is studied numerically. The frosting experiments are simulated via a multiphase Eulerian-Eulerian frosting model developed in ANSYS FLUENT, and the results are compared to predict frosting phenomena on these surfaces. It is found that AM surfaces significantly affect frost formation and heat transfer characteristics.Auburn Universit
Transfer öğrenme görsel özellikleri ile NFT satış özellikleri ve fiyat tahmini.
No full textNon-fungible tokens~(NFTs) are unique digital assets whose possession is defined over a blockchain. NFTs can represent multiple distinct objects such as art, images, videos, etc. NFTs are almost always traded by using cryptocurrencies such as Ethereum and blockchains are utilized to encode them. There was a recent surge of interest in trading them which makes them another type of alternative investment. While the existing research predominantly focuses on the technical aspects of NFTs, limited attention has been given to predicting their prices. The inherent volatility of NFT prices, attributed to factors such as over-speculation, liquidity constraints, rarity, subjectivity, and market volatility, presents challenges for accurate price predictions. For such analysis and forecasting, machine learning methods offer a robust solution framework. Here, we focus on two related prediction problems over NFTs: Predicting NFTs' sale price, and inferring whether a given NFT will participate in a secondary sale. We analyze and learn the visual characteristics of NFTs by deep pre-trained models and combine such visual knowledge with additional important non-visual attributes such as the sale history, trader behavior, seller's and buyer's centralities in the trading network, and collection's resale probability, and we assess the reliability of these features. We categorize input NFTs into six categories based on their characteristic features: Art, Collectibles, Games, Metaverse, Utility, and Others. We train several different machine learning methods on these attributes to answer these two questions. Across detailed experiments, we found visual attributes obtained from deep pre-trained models to increase the prediction performance in all cases, even though the pre-trained model giving the optimal result may change depending on the problem type. In general, none of the learning algorithms consistently outperformed the rest of them across all categories. Our code is publicly available at https://github.com/seferlab/deep_nft.Nitelikli Fikri Tapu'lar (NFT'ler); temel olarak resimler, sanat eserleri, hareketli görseller ve mizahi tasarımlar gibi görselleştirilebilecek herhangi bir nesneyi temsil eden, mülkiyeti blok zinciri üzerinde tanımlanmış dijital varlıklardır. Genellikle Ethereum gibi kripto para birimleri ile çevrimiçi alınıp satılabilirler. NFT'lere ve ticaretine olan ilginin artışı, NFT'leri alternatif bir yatırım aracı haline getirmektedir. Ancak mevcut araştırma ve çalışmalar genellikle NFT'lerin teknik yönlerine odaklanmıştır ve fiyat tahminine yönelik çalışmalar sınırlı kalmıştır. NFT piyasasının karmaşık yapısı, spekülasyon ve manipülasyona açık olması, likidite eksikliği, nadirlik, öznellik ve volatilite gibi faktörler fiyat tahminini oldukça zor hale getirmektedir. Bu bağlamda, analiz ve tahminler için makine öğrenimi güçlü bir çözüm sunmaktadır. Bu çalışmada iki tahmin problemine odaklanılmıştır: NFT'lerin satış fiyatı ve ikincil satış olasılığı. Önceden eğitilmiş derin öğrenme algoritmaları kullanılarak NFT'lerin görsel özellikleri analiz edilmiş, bu görsel bilgi ile satış geçmişi, tüccar davranışı ve ikincil satışların fiyat üzerindeki etkisi de analiz ve öğrenme sürecine dahil edilmiş ve bu özelliklerin güvenilirlikleri değerlendirilmiştir. NFT'leri sanat, koleksiyon, oyun, metaverse, işlevsellik ve diğer olmak üzere 6 kategoriye ayırarak makine öğrenimi algoritmalarıyla satışlar modellenmiş ve bu problemlere cevap aranmıştır. Denemeler sonucunda, görsel özelliklerin tahmin performansını tüm durumlarda artırdığı, optimum performansı gösteren modelin ise problem tipine göre değiştiği görülmüştür. Genel olarak, öğrenme algoritmalarının hiçbiri tüm kategorilerde sürekli olarak diğerlerinden daha iyi performans gösterememiştir. Kodlarımız, https://github.com/seferlab/deep_nft adresinde herkese açık olarak sunulmuştur
Augmenting reservoirs with higher order terms for resource efficient learning
No full textReservoir computing provides an attractive alternative for time series data representation due to their gradientfree learning and energy efficient operation. In a 'reservoir computer' (RC), the reservoir is a random recurrent neural network that forms a high dimensional non-linear dynamical system which can be mapped to a desired sequence or dynamical system through linear regression. Although, the non-linearity of the reservoir gives its power, there is no guarantee that it is the most suitable for the task at hand. To this end, in this study, we propose to amplify the effectiveness of reservoirs by augmenting them with higher order terms computed based on reservoir unit activations. We name this model as Higher-orderaugmented Reservoir Computer (ha-RC). We test the efficacy of ha-RCs by using two types of tasks: time series representation and long-term prediction, i.e. learning a dynamical system. Our experiments show that with the proposed model, a given learning accuracy can be achieved with significantly less memory resource compared to the baseline of standard Reservoir Computer (sRC). This becomes possible as ha-RC can handle complex learning problems with much smaller reservoirs compared to sRC models. This memory efficiency directly translates to energy efficiency as less number of arithmetic operations are needed with ha-RCs to reach the same level of accuracy compared to sRCs. These points make our proposed model ideal for hardware implementation and edge computing.New Energy and Industrial Technology Development Organization ; Japan Science and Technology Agency ; Japan Society for the Promotion of Science ; Core Research for Evolutional Science and Technolog
Autobiographical memory of blind and sighted early teenagers: Memory accessibility, episodicity and phenomenology
No full textVisual system is crucial to autobiographical memory. Research tended to show that blind adults may compensate for the loss of visual information in retrieval of their autobiographical memories. Much less is known about how blind children's autobiographical memory develops in the absence of visual information. Using cue-word methodology, 36 sighted and 33 blind early teenagers were asked to recall memories and subsequently rated phenomenological qualities of their memories. Retrieval latency, the number of prompts provided, episodic and non-episodic details reported for each memory were coded. In terms of memory accessibility, the blind group recalled comparable number of memories with comparable latency to retrieve memories, but they needed more prompting. Blind participants recalled similar number of episodic details; however, they reported more extraneous details, decreasing specificity. Blind early teenagers reported higher auditory imagery, a propensity to remember events from the first-person perspective, and a tendency to remember events as coherent stories
Transformer fonksiyon yaklaşımcısını kullanarak derin Q-Ağı tabanlı kripto para yatırım stratejileri.
No full textCryptocurrencies have recently started to become more apparent as an alternative investment class, as they are mainly decentralized and transparent. Since cryptocurrencies have unique features and are more volatile, we need to develop novel approaches for more accurate price prediction and more profitable investment strategies. Deep learning methods including deep reinforcement learning and transformers have been attracting remarkably more attention. Here, we come up with Reincrypt which applies Deep Q-Network (DQN) with a vision transformer (ViT) neural network function approximation, where we transform one-dimensional time-series data to 2D grayscale image-like data by using a set of important technical analysis indicators. Such transformed data will be taken as input in our framework to predict various cryptocurrency prices. To stabilize the learning process of the model and enhance its performance, we employ techniques like parameter freezing and experience replay. Our detailed findings show that the proposed method is successful in forecasting changes in cryptocurrency prices. Overall, our method yields profit in different cryptocurrencies, not just Bitcoin. When we analyze the portfolios formed as a result of our method's output, we obtain approximately 0.5% return per transaction before considering transaction costs for the cryptocurrencies. Additionally, we can predict future cryptocurrency prices even when we train our method on a different cryptocurrency. In this case, we can train our method on highly-traded and liquid cryptocurrencies such as Bitcoin, and test its performance on relatively less liquid alternative cryptocurrencies. According to these results, deep learning-based cryptocurrency price prediction methods could be utilized for alternative cryptocurrencies, even though we have fewer data for these assets.Kripto paralar, çoğunlukla merkeziyetsiz ve şeffaf oldukları için alternatif bir yatırım sınıfı olarak son zamanlarda daha fazla öne çıkmaya başladı. Kripto paraların benzersiz özelliklere sahip olmaları ve daha oynak olmaları nedeniyle, daha doğru fiyat tahmini ve daha karlı yatırım stratejileri geliştirmek için yeni yaklaşımlar geliştirmemiz gerekiyor. Derin öğrenme yöntemleri arasında özellikle derin takviye öğrenmesi ve dönüştürücüler dikkat çekici derecede daha fazla ilgi görmeye başladı. Burada, bir Derin Q-Ağı (DQN) ile bir görüş dönüştürücü (ViT) sinir ağı fonksiyon yaklaşımını uygulayan Reincrypt ile karşınıza çıkıyoruz. Burada, önemli teknik analiz göstergelerinden oluşan bir set kullanarak, tek boyutlu zaman serisi verilerini 2D gri tonlamalı görüntü benzeri verilere dönüştürüyoruz. Böyle dönüştürülmüş veriler, çeşitli kripto para fiyat hareketlerini tahmin etmek için çerçevemizde girdi olarak alınacak. Modelin öğrenme sürecini stabilize etmek ve performansını artırmak için parametre dondurma ve deneyim tekrarı gibi teknikler kullanıyoruz. Detaylı bulgularımız, önerilen yöntemin kripto para fiyatlarındaki değişiklikleri tahmin etmede başarılı olduğunu gösteriyor. Genel olarak, yöntemimiz sadece Bitcoin değil, farklı kripto paralarda da kar sağlıyor. Yöntemimizin çıktısı sonucunda oluşturulan portföyleri analiz ettiğimizde, işlem başına işlem maliyetlerini dikkate almadan yaklaşık %0.5 getiri elde ediyoruz. Ayrıca, yöntemimizi farklı bir kripto para üzerinde eğittiğimizde gelecekteki kripto para fiyat hareketlerini tahmin edebildiğimizi görüyoruz. Bu durumda, yöntemimizi Bitcoin gibi yüksek işlem hacmine ve likiditeye sahip kripto paralar üzerinde eğitip, performansını nispeten daha az likit alternatif kripto paralarda test edebiliriz. Bu sonuçlara göre, derin öğrenme tabanlı kripto para fiyat hareketi tahmin yöntemleri, bu varlıklar için daha az veriye sahip olsak bile alternatif kripto paralar için kullanılabilir
Mukarnasın tarihi Türk Hamamı’nın günışığı performansına etkisi
No full textThe intangible cultural heritage features of historical buildings became an important issue after the adoption of UNESCO’s Safeguarding of Intangible Heritage. In addition to all their other traditional features, the atmosphere of the interior of Turkish baths (hammams), which includes daylight and steam, can be regarded as their intangible cultural heritage feature. Furthermore, the interrelation between daylight and the geometric features of these baths is a crucial parameter in creating this atmosphere.
Therefore, this study focused on the interaction between daylight intake and the muqarnas of traditional Turkish baths. The purpose of this study is to measure the effect of muqarnas decorations on daylight intake in the interior space of historical Turkish baths through simulations. Also, a model that can optimize the muqarnas geometry according to the desired amount of annual daylight intake in the space was proposed. A section of the bath with muqarnas in its dome transition element was selected for daylighting analysis. After the simulation setups were completed, daylighting analysis simulations were conducted on the three-dimensional models of the Turkish bath, both with and without muqarnas. According to the simulation results, muqarnas were observed to have a slight effect
on daylight intake. The proposed optimization model emphasizes the relationship between the amount of daylighting in the interior and the muqarnas. The optimization experiments demonstrated that daylighting performance can be enhanced with the help of muqarnas decorations.Tarihi yapıların somut olmayan kültürel miras özellikleri, UNESCO’nun Somut Olmayan Mirasın Korunması kararının kabul edilmesinden sonra önemli bir konu haline geldi. Türk hamamlarının diğer tüm geleneksel özelliklerinin yanı sıra iç mekanlarının günışığını ve buharı da içeren atmosferi, onların somut olmayan kültürel miras özelliği olarak kabul edilebilir. Ayrıca günışığı ile hamamların geometrik özellikleri arasındaki ilişki de bu atmosferin yaratılmasında etkili bir parametre olarak sıralanabilir. Bu nedenle bu çalışmada geleneksel Türk hamamlarının günışığı alımı ile mukarnasları arasındaki etkileşime odaklanılmıştır. Bu çalışmanın amacı, tarihi Türk hamamı iç mekanındaki mukarnas süslemelerin günışığı alımına etkisini simülasyonlar yoluyla ölçmektir. Ayrıca bu çalışmada mekanda yıllık olarak istenilen günışığı miktarına göre mukarnas geometrisini optimize edebilecek bir model önerilmiştir. Hamamın kubbe geçiş elemanı mukarnaslı olan bir odası günışığı analizi için seçilmiştir. Simülasyon kurulumları yapıldıktan sonra mukarnaslı ve mukarnassız Türk hamamının 3 boyutlu modelleri üzerinde günışığı analiz simülasyonları yapılmıştır. Simülasyon sonuçlarına göre, mukarnasın günışığı alımına az da olsa etki ettiği görülmüştür. Önerilen optimizasyon modeli ile iç mekandaki günışığı miktarı ile mukarnaslar arasındaki ilişki vurgulanmıştır. Optimizasyon deneyleri, mukarnas süslemeleri yardımıyla doğal aydınlatma performansının artırılabileceğini göstermiştir.Publisher versio
Narrating the layers of history: Exploring temporal changes and practical realities in school building renovations
No full textThis contribution explores architectural renovations of selected school buildings in Amsterdam from various periods within the Modern Era, each designed by different architects. It entails a thorough investigation of interventions spanning from initial design and construction up to 2010, categorizing them by shared themes and distinctive features. The authenticity and unique attributes of each project are central, influencing the intervention strategies implemented. Emphasis is placed on employing a visual research tool to deeply understand the various interventions shaping the historical narrative. Interventions - harmonizing heritage preservation with contemporary renewal imperatives - are chronologically sequenced, forming the core of the tool. Methodologically, the research utilizes Amsterdam municipality’s digitized data on all buildings until 2010. The selection prioritizes school buildings designated as national or municipal monuments, anticipating access to comprehensive data through official publications and secondary sources. Through a detailed historical analysis, individualized historical trajectories for each building are constructed on a timeline, highlighting technological integrations while uncovering real-life scenarios and practical necessities. The timeline, more than a chronological record, becomes an intricate storyline and contextual history. ‘Breaking points’, marking significant interventions, are crucial for understanding transformative changes impacting perception, utilization, and adaptation to contemporary requirements. Integrating textual materials, action descriptions, and analyses of construction documents and technical drawings, the timeline provides a comprehensive representation of the buildings’ evolutionary journey. In conclusion, the study proposes a multi-layered, time-adjusted monitoring dashboard to track architectural renovations. The devised flexible timeline functions as a record-keeping instrument, a streamlined information tagging system, and an integrated visual overview. This chronological data registry synthesizes different information layers within a broader historical context, thus assisting the exploration, structuring, and understanding of the complex building transformations. By complementing the documentation, archiving, and analytical processes, it could potentially serve as a forecasting tool to regulate in future actions in heritage management and planning
A systematic mapping on software testing for blockchains
No full textThe purpose of this study is to identify and classify studies published on software testing techniques applied to blockchain systems. Previously published reviews in related areas have a narrow focus and/or do not follow a systematic review protocol. We conducted a systematic mapping based on an initial selection of 1025 studies. A rigorous selection process resulted in a final pool of 17 primary studies. These studies are categorized with respect to the employed testing methods, considered quality attributes, and functionality. We observe that most of the publications focus on testing functional correctness or security, whereas the testing of runtime performance attracts less attention. Existing approaches mostly employ fuzz testing or mutation testing. Search-based testing is usually combined with these techniques. The application of model-based testing is rare. The adaptability of fuzz testing and model-based testing techniques to changing blockchain platforms and languages remains a concern. On the other hand, performance and scalability issues are noted for search-based techniques and mutation testing. The use and integration of multiple testing techniques also stand out as a viable research direction.TÜBİTAKPublisher versio