Opin vísindi

 

Nýlega bætt við

Verk
Modelling metabolomic changes in human bone marrow derived mesenchymal stem cells during oesteogenic differentiation via genome scale metabolic models to further their use in regenerative medicine
(Reykjavík University, 2021) Sigmarsdottir, Thora Bjorg; Sigurjónsson, Ólafur E.
In recent years the fields of regenerative and translational medicine have become the subjects of significantly growing interest due to their offer of previously unimaginable therapeutics. Within these fields are several novel tools believed to hold the keys to furthering existing and new developments and one of those tools is human mesenchymal stem cells. One of the applications hMSCs have been studied for is enhanced osteogenic regeneration or reconstruction of new bone tissue. Although various studies have been performed and some strides been made towards a plausible clinical application there is still lot left to be discovered. A methodical studying and relatively detailed in silico genome scale metabolic modelling occurring changes and the accompanying metabolic phenotypes could provide a means to fill in the existing knowledge gaps (from the protein level all the way to the genomic level) and, additionally, a means to perform hypotheses testing with a significant reduction when it comes to the accompanying cost. The objective of this thesis was to study the metabolomic changes in hMSCs during osteogenic differentiation using original transcriptomic, intracellular and extracellular metabolomic data in order try and define possible metabolic stages over the course of the differentiation and use genome scale network reconstruction to create in silico models. In the first part of this work extracellular and intracellular data were used to define possible stages to osteogenic differentiation and hypothesise which pathways may characterise the different metabolic phenotypes. Three stages were suggested based on the data and labelled intracellular metabolomics indicated a decrease in glycolytic dependencies throughout the differentiation period with an increase in mitochondria related energy producing functions as the osteogenesis progressed. This will help focus specific time points of interest and relevance when it comes to mapping the significant metabolomic changes. In the second part of this work extracellular metabolomic data collected from BM-hMSCs during proliferation, adipogenic and osteogenic differentiation was used along with experimentally specific data to create three directly comparable genome scale metabolic models, two of which have no comparable predecessors, for those cell lineages during the first 7 days of cell culture. Models were biologically feasible and showed all lineage specific characteristic reactions as active. By analysis and comparison, the various enriched subsystems and pathways most significant for each lineage were found. Results were varied for proliferating cells, which matches that they have to synthesise various metabolites and substances to expand, whilst fatty acid oxidation and fatty acid synthesis was most prominent in adipogenesis with fatty acid oxidation as well as transport reactions characteristic for osteogenesis. These models can be used for model-driven experimental design to engineer osteogenesis and, with modifications, to create disease model for osteoporosis. In the third part presented we summarized the various characteristics and possibilities that lie in using MSCs as a tool in tissue engineering and regenerative medicine and how, via implementation of genome scale metabolic model reconstruction, their possibilities could possibly be taken much further in a faster, more methodical manner while reducing the related experimental cost. To summarise, this work has provided real strides when it comes to closing the existing gap regarding metabolomic changes during differentiation of hMSCs as well as providing novel tools that can be used to make further studies more efficient and cost effective.
Verk
„Riddarinn hallast við brotinn brand, bleik er hans unga kinn“: Um birtingarmyndir berkla í skáldskap
(Hugvísindastofnun Háskóla Íslands, 2025-09-30) Egilsson, Sveinn Yngvi; Íslensku- og menningardeild (HÍ); Faculty of Icelandic and Comparative Cultural Studies (UI); Hugvísindasvið (HÍ); School of Humanities (UI)
Í greininni er fjallað um skáldskap sem tengist berklum. Hún hefst á almennum inngangi um sögu berkla og menningarlega þýðingu þeirra sem „rómantísks sjúkdóms“. Síðan eru dregnar saman helstu niðurstöður erlendra rannsókna á berklabókmenntum. Susan Sontag og Katherine Byrne hafa bent á skáldskapareinkenni á borð við blómamál, lit og litleysi (roða, fölva), hita og kulda, hernaðarmál (bardaga, berkla sem óvin) og persónugervingar berkla eða dauðans (vofa, svipur, beinagrind). Auk þess hafa þær sýnt að í skáldskapnum birtist oft sterkar andstæður eins og lífsþorsti og dauðaþrá, fegurð og ljótleiki, ást og hatur eða beiskja, ástríður og deyfð, sköpunarkraftur og lamandi depurð, virkni og þolandaháttur, líkami og andi. Í greininni er leitað svara við þeirri spurningu hvort hliðstæð einkenni megi finna í íslenskri ljóðagerð. Könnunin beinist sérstaklega að völdum kvæðum eftir skáldin Jóhann Gunnar Sigurðsson (1882–1906), Stefán Sigurðsson frá Hvítadal (1887–1933) og Guðfinnu Jónsdóttur frá Hömrum (1899–1946). Einnig er hugað að þeim hugmyndaklasa sem tengist sjúkdómum eins og berklum og getur falið í sér æsku, fegurð, gáfur, hnignun, depurð og dæmda ást eða ill örlög. Helstu niðurstöðurnar eru þær að í ljóðum þessara þriggja skálda, sem öll dóu úr berklum, megi greina hliðstæð einkenni og bent hefur verið á í erlendum berklaskáldskap. Áðurnefndar andstæður setja sterkan svip á kvæði þeirra allra. Mikið ber á myndmáli kulda og þrá eftir hlýju og birtu. Haust- og vetrarlandslag myndar þá gjarnan andstæðu við vor- og sumarmyndir. Í kvæðum Jóhanns Gunnars birtast veikindin oft í myndhverfingum og ummyndunum eins og þekkt er úr kvæðum eftir erlend skáld eins og John Keats (1795–1821) sem berklar drógu til dauða. Þeir yrkja báðir um föla riddara sem eru milli heims og helju (kvæðin La Belle Dame sans Merci og Í val). Jóhann Gunnar sækir stef og stíl í þjóðkvæði og lætur ljóðrænan óhugnað þeirra endurspegla vanlíðan og ótta þeirra sem eru alvarlega veik. Stefán frá Hvítadal yrkir á beinskeyttari hátt þó að hann beiti líka myndmáli til að lýsa þeirri líðan og tilfinningum sem fylgja því að vera haldinn lífshættulegum sjúkdómi. Kvæði Stefáns einkennast af hispurslausri tjáningu og tilfinningasveiflum sem lýsa von um bata og ótta við dauðann. Guðfinna frá Hömrum skapar ákveðinn ljóðheim sem helgast af fegurð, samhljómi og birtu. Tónlist setur sterkan svip á skáldskapinn og myndar eins konar hljómgrunn í lýsingum á náttúrunni og tilfinningum. Guðfinna yrkir um dulúðuga tóna og hljóma sem óma í sköpunarverkinu og tengja má hugmyndinni um „hljómlist hvolfanna“ (lat. Musica universalis, e. Music of the spheres). Keats yrkir á svipaðan hátt í Ode on a Grecian Urn og skapar í ljóðum sínum heim fegurðar og samhljóms. Nálægðin við dauðann og undirliggjandi depurð koma víða fram í ljóðum Guðfinnu og geta tekið á sig tónlistarmyndir. Þannig má greina ýmis einkenni berklabókmennta í ljóðum Guðfinnu en í samanburði við skáldskap Jóhanns Gunnars og Stefáns er ljóðheimur hennar fagurfræðilegri og hljómrænni. Í lok greinarinnar er hvatt til frekari rannsókna á íslenskum bókmenntum sem tengjast berklum.
Verk
Dual-Polarized Super-Wideband Antenna with Artificial Materials-Enhanced Gain for 5G Millimeter-Wave and Remote Sensing Applications
(2026) Esmail, Bashar A.F.; Isleifson, Dustin; Koziel, Slawomir; Pietrenko-Dabrowska, Anna; Department of Engineering
This paper presents a high-performance dual-polarized super-wideband (SWB) antenna that employs optimized slots and an Artificial Material (AM) array to achieve remarkable enhancements in both bandwidth and gain. The proposed design features a compact Vivaldi configuration with integrated arrow-shaped slots, enabling SWB operation while maintaining a low-profile size of 0.66× 0.49× 0.01λ3. A constrained trust-region gradient-based optimization is used to refine the slot geometry and antenna parameters, achieving an operating range from 8.4 to 42.8 GHz with a stable gain of approximately 10 dBi at frequencies above 17 GHz. The incorporation of AMs further enhances the gain and extends the bandwidth. The AM antenna, with dimensions of 0.94× 0.49× 0.01λ3 , achieves peak gains of 13.3 dBi at 24 and 38 GHz and covers a wide operating range of 8.3-45 GHz. Dual polarization is realized by employing two orthogonally oriented Vivaldi elements, resulting in a high polarization isolation exceeding 25 dB across the entire operating bandwidth. The experimental results demonstrate excellent agreement with the simulation-based findings. Featuring SWB performance, high gain, strong polarization isolation, polarization diversity, and a compact low-profile structure, the proposed antenna is well suited for 5G millimeter-wave (MMW) and satellite remote sensing applications.
Verk
Dual adversarial attacks : Fooling humans and classifiers
(2023-06) Schneider, Johannes; Apruzzese, Giovanni; Department of Computer Science
Adversarial samples mostly aim at fooling machine learning (ML) models. They often involve minor pixel-based perturbations that are imperceptible to human observers. In this work, adversarial samples should fool both humans and ML models, which is important in two-stage decision processes. We perform changes on a higher abstraction level so that a target sample exhibits properties of a desired sample. Technically, we contribute by deriving a regularization scheme for autoencoders incorporating a classifier loss for smoothly interpolating between wildly different samples. The realism and effectiveness of generated samples are confirmed with a user study and other evaluations. Our experiments consider neural networks of four architectures, assessed on MNIST, FashionMNIST, QuickDraw and CIFAR-10. Results show that our scheme leads to superior performance compared to existing interpolation techniques: on average, other methods have an 11% higher failure rate when producing a sample that is of any of two interpolated classes. Furthermore, our attacks work in both white- and black-box settings.
Verk
DReLAB - Deep REinforcement Learning Adversarial Botnet : A benchmark dataset for adversarial attacks against botnet Intrusion Detection Systems
(2021-02) Venturi, Andrea; Apruzzese, Giovanni; Andreolini, Mauro; Colajanni, Michele; Marchetti, Mirco; Department of Computer Science
We present the first dataset that aims to serve as a benchmark to validate the resilience of botnet detectors against adversarial attacks. This dataset includes realistic adversarial samples that are generated by leveraging two widely used Deep Reinforcement Learning (DRL) techniques. These adversarial samples are proved to evade state of the art detectors based on Machine- and Deep-Learning algorithms. The initial corpus of malicious samples consists of network flows belonging to different botnet families presented in three public datasets containing real enterprise network traffic. We use these datasets to devise detectors capable of achieving state-of-the-art performance. We then train two DRL agents, based on Double Deep Q-Network and Deep Sarsa, to generate realistic adversarial samples: the goal is achieving misclassifications by performing small modifications to the initial malicious samples. These alterations involve the features that can be more realistically altered by an expert attacker, and do not compromise the underlying malicious logic of the original samples. Our dataset represents an important contribution to the cybersecurity research community as it is the first including thousands of automatically generated adversarial samples that are able to thwart state of the art classifiers with a high evasion rate. The adversarial samples are grouped by malware variant and provided in a CSV file format. Researchers can validate their defensive proposals by testing their detectors against the adversarial samples of the proposed dataset. Moreover, the analysis of these samples can pave the way to a deeper comprehension of adversarial attacks and to some sort of explainability of machine learning defensive algorithms. They can also support the definition of novel effective defensive techniques.

Flokkar í Opnum vísindum

Veldu flokk til að skoða.

Niðurstöður 1 - 9 af 9

Öll gögn í Opnum vísindum eru vernduð af ákvæðum höfundalaga og með öllum réttindum áskildum, nema annað sé tekið fram.