OuluREPO University of Oulu repository
Not a member yet
54977 research outputs found
Sort by
Impact of MAC Unit Design Architectures and Their Applications in Modern Computing: An In-Depth Review
No full textAbstract
A fundamental computational block essential for computing devices is the multiplier-accumulator (MAC) unit. A MAC unit repeatedly performs a multiply and add operation commonly used in many data processing and artificial intelligence (AI) applications. This article comprehensively analyzes MAC unit architectures, focusing on high-throughput configurations such as systolic arrays and tree-based multipliers integrated into AI, high-performance computing (HPC), and digital signal processing (DSP) hardware. It traces their evolution from serial and pipelined designs to specialized precision formats (fixed and floating point) used in modern workloads. We explore modern design concepts, including approximate MACs utilizing truncated arithmetic, sparsity-aware designs that bypass zero-value operands, and units optimized for binary neural networks. The article also examines the transformative impact of In-Memory Computing (IMC) MACs, which leverage crossbar arrays to mitigate data movement bottlenecks. Furthermore, we contextualize these advancements within leading hardware accelerators, such as Google Tensor Processing Unit (TPUs), field programmable gate arrays (FPGAs), and NVIDIA Tensor cores. We identify persistent challenges, including power constraints and reliability in nanoscale technologies, and also outline future trends such as photonic MACs and adaptive bit-width scaling, which can serve as a guide for future innovation in this vital field.Abstract
A fundamental computational block essential for computing devices is the multiplier-accumulator (MAC) unit. A MAC unit repeatedly performs a multiply and add operation commonly used in many data processing and artificial intelligence (AI) applications. This article comprehensively analyzes MAC unit architectures, focusing on high-throughput configurations such as systolic arrays and tree-based multipliers integrated into AI, high-performance computing (HPC), and digital signal processing (DSP) hardware. It traces their evolution from serial and pipelined designs to specialized precision formats (fixed and floating point) used in modern workloads. We explore modern design concepts, including approximate MACs utilizing truncated arithmetic, sparsity-aware designs that bypass zero-value operands, and units optimized for binary neural networks. The article also examines the transformative impact of In-Memory Computing (IMC) MACs, which leverage crossbar arrays to mitigate data movement bottlenecks. Furthermore, we contextualize these advancements within leading hardware accelerators, such as Google Tensor Processing Unit (TPUs), field programmable gate arrays (FPGAs), and NVIDIA Tensor cores. We identify persistent challenges, including power constraints and reliability in nanoscale technologies, and also outline future trends such as photonic MACs and adaptive bit-width scaling, which can serve as a guide for future innovation in this vital field
Generalised level anticrossings explain improved 19F SABRE hyperpolarisation under oscillating magnetic fields
No full textAbstract
Signal amplification by reversible exchange (SABRE) is a parahydrogen-based hyperpolarisation technique that significantly enhances nuclear magnetic resonance signals without the use of expensive hardware. While conventional SABRE relies on a static polarisation transfer field set near the level anticrossing (LAC) condition, recent work has shown that oscillating fields can substantially boost the hyperpolarisation levels. Here, we develop a new theoretical model that generalises the LAC condition to account for the oscillating polarisation transfer fields, thereby explaining the spin dynamics of SABRE under such conditions. We use the generalised LAC condition to optimise the oscillating fields for maximum polarisation transfer to 19F and show that the scalar relaxation of the second kind can be simultaneously suppressed. Large-scale spin dynamics simulations and experiments show that this leads to enhanced 19F hyperpolarisation compared to conventional SABRE, with a 79% improvement observed experimentally. This work demonstrates a generalisable strategy for improving the efficiency of SABRE, advancing its potential for various applications, such as in biomedicine.Abstract
Signal amplification by reversible exchange (SABRE) is a parahydrogen-based hyperpolarisation technique that significantly enhances nuclear magnetic resonance signals without the use of expensive hardware. While conventional SABRE relies on a static polarisation transfer field set near the level anticrossing (LAC) condition, recent work has shown that oscillating fields can substantially boost the hyperpolarisation levels. Here, we develop a new theoretical model that generalises the LAC condition to account for the oscillating polarisation transfer fields, thereby explaining the spin dynamics of SABRE under such conditions. We use the generalised LAC condition to optimise the oscillating fields for maximum polarisation transfer to 19F and show that the scalar relaxation of the second kind can be simultaneously suppressed. Large-scale spin dynamics simulations and experiments show that this leads to enhanced 19F hyperpolarisation compared to conventional SABRE, with a 79% improvement observed experimentally. This work demonstrates a generalisable strategy for improving the efficiency of SABRE, advancing its potential for various applications, such as in biomedicine
'Honestly, i could lose a few hours on PubMed quite easily': breaking down the reading barriers of medical literature with an LLM-powered browser extension
No full textAbstract
Scientific literature is a primary source of validated information. Sometimes, however, the complicated field-specific language and terminology act as barriers to understanding the insights. This is also true for the societally significant medical field. With the recent rapid development of Large Language Models (LLMs), Artificial Intelligence (AI) can be useful in mitigating these issues. We implemented a browser extension (SimpliMed) to simplify PubMed abstracts with GPT-4 (implemented by ChatGPT API) to make medical literature more accessible to wider audiences and tested it with a user study. Our results indicate that SimpliMed was effective in simplifying medical article abstracts and helped support understanding, discovering new knowledge, and developing critical thinking skills. Our findings contribute to the understanding of tools with LLMs to simplify medical literature. We shed light on its multiple implications that extend beyond the immediate scope of medical literature simplification, including fighting misinformation and future education.Abstract
Scientific literature is a primary source of validated information. Sometimes, however, the complicated field-specific language and terminology act as barriers to understanding the insights. This is also true for the societally significant medical field. With the recent rapid development of Large Language Models (LLMs), Artificial Intelligence (AI) can be useful in mitigating these issues. We implemented a browser extension (SimpliMed) to simplify PubMed abstracts with GPT-4 (implemented by ChatGPT API) to make medical literature more accessible to wider audiences and tested it with a user study. Our results indicate that SimpliMed was effective in simplifying medical article abstracts and helped support understanding, discovering new knowledge, and developing critical thinking skills. Our findings contribute to the understanding of tools with LLMs to simplify medical literature. We shed light on its multiple implications that extend beyond the immediate scope of medical literature simplification, including fighting misinformation and future education
Genetic relatedness mattered in the co-burial ritual of Neolithic hunter-gatherers
No full textAbstract
Kin relations among past societies can offer valuable information about the social dynamics of the population. Genetic data from prehistoric human remains can reveal genetic relatedness, and when combined with archaeological information, shed light on social factors shaping ancient communities. However, accessing such information on ancient hunter–gatherer societies has been challenging owing to the scarcity of temporally overlapping multi-burial sites. Here, we focused on the Pitted Ware Culture (PWC) cemetery from Ajvide (Gotland, Sweden), one of Stone Age Europe’s largest and best-preserved hunter–gatherer burial grounds of the European Stone Age. We generated new genomic data from 10 individuals, primarily from co-burial contexts, and combined these with published genomes from 24 individuals across four PWC sites on Gotland. The genetic analyses revealed dual ancestry of the Gotlandic PWC, showing approximately 80% ancestry associated with earlier Mesolithic hunter–gatherer groups and 20% with farmer groups. We also identified close genetic relatives between the different studied PWC sites on Gotland, indicating mixing of the groups. All individuals buried together were closely related to one another, including first-, second- and third-degree relatives, and showed significantly elevated genetic relatedness. This demonstrates that genetic relatedness played a defining role in the co-burial ritual and extended beyond first-degree relatives.Abstract
Kin relations among past societies can offer valuable information about the social dynamics of the population. Genetic data from prehistoric human remains can reveal genetic relatedness, and when combined with archaeological information, shed light on social factors shaping ancient communities. However, accessing such information on ancient hunter–gatherer societies has been challenging owing to the scarcity of temporally overlapping multi-burial sites. Here, we focused on the Pitted Ware Culture (PWC) cemetery from Ajvide (Gotland, Sweden), one of Stone Age Europe’s largest and best-preserved hunter–gatherer burial grounds of the European Stone Age. We generated new genomic data from 10 individuals, primarily from co-burial contexts, and combined these with published genomes from 24 individuals across four PWC sites on Gotland. The genetic analyses revealed dual ancestry of the Gotlandic PWC, showing approximately 80% ancestry associated with earlier Mesolithic hunter–gatherer groups and 20% with farmer groups. We also identified close genetic relatives between the different studied PWC sites on Gotland, indicating mixing of the groups. All individuals buried together were closely related to one another, including first-, second- and third-degree relatives, and showed significantly elevated genetic relatedness. This demonstrates that genetic relatedness played a defining role in the co-burial ritual and extended beyond first-degree relatives
A generative adversarial network-based methodology in systematic machine design at concept phase for solutions
No full textAbstract
Artificial intelligence (AI) has made a significant impact in engineering through its data analysis and interpretation capabilities. In systematic machine design, AI tools can enhance the design process, from improving quality analysis in computer-aided design (CAD) drawings to generating three-dimensional models and inverse design structures. This research proposes a methodology for using generative AI in the conceptual design phase, based on the StyleGAN2ADA architecture, an image generation model designed for high-quality synthesis with limited training data. The method aims to preserve trademark or otherwise recognizable features in the design while exploring new combinations of desired design attributes. A case study demonstrates the approach by blending visual features from cars and shoes using a combined dataset of 3667 images, with StyleGAN2-ADA trained to an optimal point of ∼4800 kimg. Across the generated samples, about 2% of generated images showed clear car–shoe feature mixing, while fewer than 2% were classified as blurred, and the full training run required on the order of 6–10 GPU hours depending on hardware configuration. These results indicate that StyleGAN2-ADA-based methods have practical potential for supporting conceptual design while also revealing limitations in mixing rate, controllability, and output quality that warrant further work.Abstract
Artificial intelligence (AI) has made a significant impact in engineering through its data analysis and interpretation capabilities. In systematic machine design, AI tools can enhance the design process, from improving quality analysis in computer-aided design (CAD) drawings to generating three-dimensional models and inverse design structures. This research proposes a methodology for using generative AI in the conceptual design phase, based on the StyleGAN2ADA architecture, an image generation model designed for high-quality synthesis with limited training data. The method aims to preserve trademark or otherwise recognizable features in the design while exploring new combinations of desired design attributes. A case study demonstrates the approach by blending visual features from cars and shoes using a combined dataset of 3667 images, with StyleGAN2-ADA trained to an optimal point of ∼4800 kimg. Across the generated samples, about 2% of generated images showed clear car–shoe feature mixing, while fewer than 2% were classified as blurred, and the full training run required on the order of 6–10 GPU hours depending on hardware configuration. These results indicate that StyleGAN2-ADA-based methods have practical potential for supporting conceptual design while also revealing limitations in mixing rate, controllability, and output quality that warrant further work
Arkkitehtuurin sukupuolittuneiden elementtien uudelleenmuotoilu : tutkielma sukupuolitietoisen arkkitehtuurin harjoittamisen demokraattisesta potentiaalista
No full textIn the current political era, gender rights are once again under discussion in a way comparable to the previous time of discussion on women’s rights beginning in the 20th century in Finland. Although in present times, basic rights have generally been guaranteed to also gender-diverse people, gender diversity has been faced with strong injustices by law, such as the compulsory sterilization requirement for confirming legal gender tcorresponding gender identity, which was only abolished in the renewed 2023’s trans law in Finland. The rights of gender-diverse people are being discussed especially now in the 2020s, and gender-related changes are also visible in the practice of architecture in Finland.
The dichotomous male-female divisions of modernist architecture have been challenged in the new millennium, and new gender-neutral spaces have been designed especially for urban environments where gender equality issues have been addressed more. The master’s thesis research examines the theme of gender through modernist architecture and raises the topic of multidisciplinary social perspective of architecture in addition. The study examines the role of social space in architectural spaces and architecture’s potential to create democratic spaces, which seek to unite people beyond gender.
The study explores the intersections of gender and architecture, utilizing literature review and ethnographic methods of spatial observation and spatial analysis with a focus on public architecture. The theoretical framework’s literature review includes a variety of materials from news, articles, books and images, connecting them with history of Finnish urbanization and modernism, construction of political rights and anthropological interpretations on gender roles and norms
Multiple full-length variants of the mitochondrial COI DNA barcode region are prevalent in north European sawflies
No full textAbstract
DNA barcoding, the use of a standard DNA fragment for species identification, has emerged as a major field of biodiversity research. The effectiveness of this approach rests on the premise that much less variation exists within species than between them. While exceptions occur, this has been demonstrated in many animal taxa where the COI gene is effective in species discrimination. Sawflies are an exception to this pattern because DNA barcodes often fail to distinguish congeneric species. Using high-throughput single-molecule DNA sequencing to recover COI sequences from thousands of sawflies, we found that single individuals often possess multiple, seemingly functional, full-length DNA barcodes (i.e., unrecognizable as nuclear pseudogenes)—a phenomenon not documented at similar prevalence in any animal taxon. While the evolutionary causes of multiple variants require further investigation, our observation is remarkable as it violates the one-barcode-one-specimen assumption. The presence of multiple variants of barcodes within individuals does not jeopardize the concept, but it introduces a complexity for species inventories based on metabarcoding. They will overestimate the species count when barcode-based operational species units are used as species proxies. Similarly, DNA barcode reference libraries must consider how best to deal with the high frequency of multiple intra-individual variants.Abstract
DNA barcoding, the use of a standard DNA fragment for species identification, has emerged as a major field of biodiversity research. The effectiveness of this approach rests on the premise that much less variation exists within species than between them. While exceptions occur, this has been demonstrated in many animal taxa where the COI gene is effective in species discrimination. Sawflies are an exception to this pattern because DNA barcodes often fail to distinguish congeneric species. Using high-throughput single-molecule DNA sequencing to recover COI sequences from thousands of sawflies, we found that single individuals often possess multiple, seemingly functional, full-length DNA barcodes (i.e., unrecognizable as nuclear pseudogenes)—a phenomenon not documented at similar prevalence in any animal taxon. While the evolutionary causes of multiple variants require further investigation, our observation is remarkable as it violates the one-barcode-one-specimen assumption. The presence of multiple variants of barcodes within individuals does not jeopardize the concept, but it introduces a complexity for species inventories based on metabarcoding. They will overestimate the species count when barcode-based operational species units are used as species proxies. Similarly, DNA barcode reference libraries must consider how best to deal with the high frequency of multiple intra-individual variants
Paper industry’s transformation to paperboard industry
No full textTyön tavoitteena on luoda katsaus paperiteollisuuden kehitykselle Suomessa sekä alan osittaisesta muuntumisesta kartongin valmistukseen. Alkuun onkin nostettu alan teollisuuden merkitystä maan taloudelle, paperikoneiden ja tuotantomäärien vähenemistä mutta myös kartongin kasvua.
Työssä luodaan kirjallisuuskatsaus paperi- ja kartonkiteollisuuden eroavaisuuksista. Tätä selvittääkseen työhön on otettu määritettäväksi tyypillisimmät merkittävimmät, Suomessa valmistetut, valmistettavat ja kulutettavat paperi- sekä kartonkilaadut, niiden ominaisuudet, koostumukset ja käyttökohteet.
Valmistuslaitteiston eroavaisuuksien havainnollistamiseksi läpi käydään paperikoneen toimintaperiaate sekä yleisimmät rakennevaihtoehdot. Tätä seuraavassa osiossa kartonkikone käydään myös lävitse osioittain siten, kuin se tyypillisesti paperin valmistuksessa käytettävästä laitteistosta eroaa.
Työn tekemiseen on käytetty apuna paperi- ja metsäteollisuuteen liittyvää kirjallisuutta, artikkeleita sekä aiheeseen liittyvää opetusmateriaalia
CREATING A COMMON GROUND FOR PROFESSIONAL DEVELOPMENT OF UNIVERSITY CHEMISTRY (STEM) LECTURERS IN EUROPE
No full textAbstract :
Today, we are faced with immense global challenges in finding sustainable equilibria between socio-economic, political, and ecological concerns. The European Chemistry Thematic Network (ECTN), the European University Association (EUA) and the European Commission are committed to sustainable improvement of the quality of university chemistry education to cope with these challenges. In this position paper, we advocate the creation of the Eurolecturer Academy (ELA), an innovative, European state of the art higher education learning platform serving academics in their continuous professional development of teaching competences and thereby supporting academics to educate students to be successful in the changing world. Within this newly established educational entity, there will be two levels of membership, Associated membership and Full membership. The ELA will not only facilitate continuous professional development of university teaching staff but will at the same time create a structure to support recognition of teaching competences of lecturers within the European dimension in teaching and learning. The certification will profit from the new 5th European Qualification Framework for micro-credentials, providing a much needed “academic currency” for the purpose of recognition of academic credentials. The ELA micro-credentials will be issued by certifying the learning outcomes of short-term learning experiences in the field of teaching and learning in higher education. The ELA will provide a micro-credentials catalogue that will address the needs for professional development of lecturers and ensure the quality of the micro-credentials through close cooperation with the internationally operating accreditation organization ASIIN (https://www.asiin.de/en/) using quality standards and valid assessment according to international best practice.Abstract :
Today, we are faced with immense global challenges in finding sustainable equilibria between socio-economic, political, and ecological concerns. The European Chemistry Thematic Network (ECTN), the European University Association (EUA) and the European Commission are committed to sustainable improvement of the quality of university chemistry education to cope with these challenges. In this position paper, we advocate the creation of the Eurolecturer Academy (ELA), an innovative, European state of the art higher education learning platform serving academics in their continuous professional development of teaching competences and thereby supporting academics to educate students to be successful in the changing world. Within this newly established educational entity, there will be two levels of membership, Associated membership and Full membership. The ELA will not only facilitate continuous professional development of university teaching staff but will at the same time create a structure to support recognition of teaching competences of lecturers within the European dimension in teaching and learning. The certification will profit from the new 5th European Qualification Framework for micro-credentials, providing a much needed “academic currency” for the purpose of recognition of academic credentials. The ELA micro-credentials will be issued by certifying the learning outcomes of short-term learning experiences in the field of teaching and learning in higher education. The ELA will provide a micro-credentials catalogue that will address the needs for professional development of lecturers and ensure the quality of the micro-credentials through close cooperation with the internationally operating accreditation organization ASIIN (https://www.asiin.de/en/) using quality standards and valid assessment according to international best practice