Septentrio Academic Publishing
Not a member yet
6415 research outputs found
Sort by
Vita Andreæ Celsii
No full textThis file is a facsimile of an obituary for Anders Celsius, published in the Acta (Proceedings) of the Uppsala-based Societas Regia Literaria et Scientiarum (Royal Society of Learning and Sciences, or Kungl. Vetenskaps-Societeten i Uppsala) for the years 1744 to 1750
"Jeg har aldri brukt nynorsk før": Undervisningskompetanse i nynorsk og bokmål
No full textTeachers who teach in primary schools in Norway are required to master both Norwegian Bokmål and Norwegian Nynorsk, the two varieties of written Norwegian. Several studies have shown that the teacher education institutions do not ensure this competence, and that the teacher students entering the profession have a low level of competence in Norwegian Nynorsk. This study examines the attitudes of teachers in further education towards Norwegian Nynorsk and the connections between attitudes and the teachers’ professional background, their written language background, and their written language practices. The basis of the study is an online survey where primary school teachers taking part in an online in-service course in Norwegian were the respondents. The article contributes insights into the attitudes that some teachers have towards Nynorsk in education, and into possible connections between the self-reported attitudes and the use of Norwegian Nynorsk in teaching. The teachers who participated in the survey attach different value to Norwegian Nynorsk and Norwegian Bokmål, and these positions are expressed when the teachers practice their profession. A main finding is that the teachers in the survey have varying degrees of competence, and thus mastery, of Norwegian Nynorsk and Bokmål, and this is expressed when they practice their profession. There is also a tendency for respondents who have a background in primary school teacher education to be somewhat more positive about Norwegian Nynorsk than those who have a background from the other educational groups represented in the survey.Det er eit krav til lærarar som underviser i skulen i Noreg at dei meistrar både nynorsk og bokmål. Fleire undersøkingar viser at lærarutdanningsinstitusjonane ikkje sikrar denne kompetansen, og at lærarstudentane som kjem ut i skulen, har for dårleg kompetanse i nynorsk. Denne studien undersøkjer haldningane lærarar under vidareutdanning har til nynorsk, og samanhengar mellom haldningar og den faglege bakgrunnen til lærarane, skriftspråkbakgrunnen deira og den skriftspråklege praksisen deira. Til grunn for studien ligg ei online spørjeundersøking der respondentane er grunnskulelærarar som var studentar på eit nettbasert Kompetanse for kvalitet-kurs i norsk studieåret 2019/2020 ved Høgskulen på Vestlandet. Kurset rekrutterer studentar frå heile landet, og det store fleirtalet av studentane definerer seg sjølve som bokmålsbrukarar. Artikkelen bidrar med innsikt i haldningar som eit utval lærarar har til nynorsk i opplæringa, og i eventuelle samanhengar mellom dei sjølvrapporterte haldningane og bruken av nynorsk i undervisninga. Eit hovudfunn er at lærarane i undersøkinga kjenner ulik grad av kompetanse, og såleis meistring, i nynorsk og bokmål, og dette kjem til uttrykk når dei utøver yrket sitt. Det er vidare ein svak tendens til at respondentane som har bakgrunn frå grunnskulelærarutdanningane, er noko meir positive til nynorsk enn dei som har bakgrunn frå dei andre utdanningsgruppene som er representerte i undersøkinga
Dense FixMatch: a simple semi-supervised learning method for pixel-wise prediction tasks
No full textWe propose Dense FixMatch, a simple method for online semi-supervised learning of dense and structured prediction tasks combining pseudo-labeling and consistency regularization via strong data augmentation. We enable the application of FixMatch in semi-supervised learning problems beyond image classification by adding a matching operation on the pseudo-labels. This allows us to still use the full strength of data augmentation pipelines, including geometric transformations. We evaluate it on semi-supervised semantic segmentation on Cityscapes and Pascal VOC with different percentages of labeled data and ablate design choices and hyper-parameters. Dense FixMatch significantly improves results compared to supervised learning using only labeled data, approaching its performance with 1/4 of the labeled samples
Identifying and Mitigating Flaws of Deep Perceptual Similarity Metrics
No full textMeasuring the similarity of images is a fundamental problem to computer vision for which no universal solution exists. While simple metrics such as the pixel-wise L2-norm have been shown to have significant flaws, they remain popular. One group of recent state-of-the-art metrics that mitigates some of those flaws are Deep Perceptual Similarity (DPS) metrics, where the similarity is evaluated as the distance in the deep features of neural networks.However, DPS metrics themselves have been less thoroughly examined for their benefits and, especially, their flaws. This work investigates the most common DPS metric, where deep features are compared by spatial position, along with metrics comparing the averaged and sorted deep features. The metrics are analyzed in-depth to understand the strengths and weaknesses of the metrics by using images designed specifically to challenge them. This work contributes with new insights into the flaws of DPS, and further suggests improvements to the metrics. An implementation of this work is available online: https://github.com/guspih/deep_perceptual_similarity_analysis
Multi-lingual agents through multi-headed neural networks
No full textThis paper considers cooperative Multi-Agent Reinforcement Learning, focusing on emergent communication in settings where multiple pairs of independent learners interact at varying frequencies. In this context, multiple distinct and incompatible languages can emerge. When an agent encounters a speaker of an alternative language, there is a requirement for a period of adaptation before they can efficiently converse. This adaptation results in the emergence of a new language and the forgetting of the previous language. In principle, this is an example of the Catastrophic Forgetting problem which can be mitigated by enabling the agents to learn and maintain multiple languages. We take inspiration from the Continual Learning literature and equip our agents with multi-headed neural networks which enable our agents to be multi-lingual. Our method is empirically validated within a referential MNIST-based communication game and is shown to be able to maintain multiple languages where existing approaches cannot
Kunnskapsoppsummering og klassifisering av tiltaket: VIP Snakk om det (2. utg.)
No full textBackground: VIP Snakk om det (formerly known as VIP-programmet) is a psycho educative intervention targeting youth in first grade at Highschool (16 -17 years of age). The intervention provides information about mental health and mental disorders, including counselling on how to cope with mental health challenges. Teacher and health personnel are trained in the intervention, and in how to detect mental health problems early to implement necessary measures through the health services.
Methods: The following databases were searched to identify articles evaluating the intervention: Embase, Medline and Psycinfo, NORART Cochrane, Cristin, NORA, SCOPUS and SweMed, NREPP, NICE and Blueprint. Information from the intervention developer has also been collected. Results from the litterature searches and other information obtained, were reviewed to identify Nordic effect studies, international systematic reviews, and other Norwegian studies of the intervention.
Results: VIP Snakk om det is evaluated in a quasi-experimental design with an intervention and a control group. Data was collected pre and post intervention, with 6-, 12- and 24-months follow-up. Effects on knowledge about mental health disorders after the intervention and follow-up were moderate. Symptoms of mental health problems are only obtained after the intervention was carried out and at one-year follow-up. Due to the chosen study design and questionnaires, it is difficult to compare the results with the results from other studies in the field. In addition, anxiety symptoms were assessed using a brief scale (5 items) with unknown psychometric properties.
Conclusion: VIP is classified at evidence level 3 – Intervention with some documentation of effect.Bakgrunn: VIP Snakk om det (tidligere VIP-programmet) er et psykoedukativt program som retter seg mot elever i videregående skoler i alderen 16 og 17 år. Programmet gir informasjon om psykisk helse og psykiske lidelser samt veiledning i hvordan man håndterer slike vansker. Lærere og helsepersonell får opplæring i programmet og hvordan man oppdager problemer hos elevene tidlig for å kunne sette i verk hjelpetiltak gjennom helsesystemet.
Metode: Denne kunnskapsoppsummeringen bygger på et systematisk litteratursøk i databasene Embase, Medline og Psykinfo, NORART, Cochrane, Cristin, NORA, SCOPUS og SweMed. Det er også søkt i internasjonale kunnskapsdatabaser samt innhentet informasjon om tiltaket fra tiltakseier. Litteratursøk og annen innhentet informasjon ble gjennomgått for å identifisere nordiske effektstudier, internasjonale oppsummeringsstudier og eventuelt andre norske studier om tiltaket.
Resultater: VIP Snakk om det er evaluert i et kvasieksperimentelt design med test og kontrollgruppe, der data ble innhentet før intervensjonen, rett etter og 6, 12 og 24 måneder etter. Effektene som gjelder kunnskap om psykiske lidelser er moderate, både rett etter deltakelse i programmet og i etterundersøkelsen. Det er en svakhet at effektene ikke er konsistent rapportert for alle måletidspunkter. Psykiske problemer er bare rapportert for tidsrommet rett etter intervensjonen til ett år etter. På grunn av valgt design og spørreskjema er det vanskelig å sammenlikne effektene med andre resultater i feltet. I tillegg måles angstsymptomer med en skala som har fem spørsmål og uavklart validitet.
Konklusjon: VIP Snakk om det klassifiseres på evidensnivå 3 – Tiltak med noe dokumentasjon på effekt
Polar cod connectivity cruise 2022: Cruise Report
No full textThe Nansen Legacy Polar Cod Connectivity cruise aimed at unravelling polar cod and capelin population connectivity and relation to the physical and chemical environment in Svalbard’s fjords. The main focus areas were Isfjorden, Kongsfjorden, Storfjorden and the South-East of Svalbard. A total of 36 stations were visited to collect of information on polar cod and capelin vertical and horizontal distribution (using echosounder, pelagic and bottom trawls), their population, life cycle and size structure (observation of maturity, otoliths reading and genetic sampling), as well as their diet (stomachs content, sampling of benthic/planktonic prey). Other studies were conducted in link with the objectives of the research foci RF2 The Human Impacts and RF3 The Living Barents Sea. The cruise was also an opportunity to visit well-known monitoring stations to complete already long time series and investigate seasonal variability.
The cruise started November 12th in Longyearbyen and ended in Tromsø on November 22nd, 2022. The first two days and a half were spent on multidisciplinary sampling in thirteen stations distributed in contrasted areas inside Isfjorden. A rockhopper gear strapping was installed after the first bottom trawl to avoid mud and rocks and kept for the whole cruise. There were some difficulties also with the grab which did not always close. It was possible to spend one day sampling Kongsfjorden, thanks to relatively good weather and low sea-ice conditions. However, the presence of cables on the seafloor and irregular and unsuitable bottom types in that fjord did not allow for bottom trawling to sample demersal and benthic communities. After approximately one day of steaming south in good weather conditions, the sampling continued in Storfjorden. However, the shallowest stations north of Storfjorden and on the way out of the fjord towards the South-East of the archipelago could not be sampled due to time constraints and difficult navigational access. Finally, five multidisciplinary stations could be sampled in the South-East of Svalbard before the two-days transit to Tromsø. Overall, most stations were visited. Polar cod largest fish were maturing, but not mature. Further analyses of the samples should unveil potential genetic differences across the fjords, and image analyses of plankton net samples will be able to detect if there were already eggs in the water column at the beginning of the dark season. There was very little signal on the echosounder during most of the cruise, but the extractive sampling revealed a diversity of communities in the different fjords
Annual Report 2014
No full textCAGE studies sub-seabed methane hydrates and related greenhouse gas release processes in Arctic environments. Enormous amounts of methane hydrate, an ice-like substance, exist in the ocean floor at high pressure and low temperature. Ocean warming makes it potentially vulnerable to melting. We know that methane released in this way acted as a powerful greenhouse gas accelerating climate warming in the geological past. We aim to examine what it could mean to our future
CAGE20-7 Cruise Report: Sediment and water column analyses around flares at Norskebanken, Hinlopen and offshore Prins Karls Forland
No full textDuring the cruise CAGE 20-7 (November 2nd to 16th 2020), we observed and investigated a large active methane seepage area south from the Yermak plateau (Northern Svalbard), in particular to characterize the water column and the sediment near known flares in Norskebanken (Geissler et al., 2016). Some flares were targeted for gas composition, DNA and bubble size distribution. The combination of the R/V Kronprins Haakon and the ROV ÆGIR 6000 from the Norwegian Marine Robotics Laboratory at the University of Bergen allowed this investigation to happen in this rough environment, while collecting invaluable data from the ROV.
The cruise may be known as: CAGE20_
CAGE16-7 Cruise Report: Deployment of observatories and water column survey in the Barents Sea and offshore Svalbard
No full textThe cruise was conducted from October 16th to 25th 2016 as part of the Centre of Excellence for Arctic Gas Hydrate, Environment and Climate (CAGE) at UiT – The Arctic University of Norway.
The main goal of the cruise was to deploy two observatories in the pingo-crater area (“Yin Yang”) in the Barents Sea, and offshore Svalbard at the same location where OS2 was deployed in June 2015 and recovered in May 2016. The exact targeted location was determined according to multibeam analysis in order to deploy in the vicinity of bubble streams.
The previously K-Landers called OS1 (black) and OS2 (grey) are now called K-Lander 1 and K-Lander 2. They contain a CH4 sensor, a CO2 sensor, a CTD, a pH meter, and a hydrophone. K-Lander 1 contains in addition a DVS (Doppler volume sampler) as the broken ADCP did not arrive on time for the deployment. K-Lander 2 contains in addition an ADCP, a transmissometer (for turbidity measurement) and an M3 (Multibeam echosounder).
The present cruise also aimed at investigating an area of extensive flares western Svalbard, particularly the shallow shelf and shelf edge. The addressed scientific topics include quantification of methane concentration in the water column, temperature and salinity (via CTD casts), echosounder and multibeam signals and current (amplitude and direction)
The cruise may be known as: CAGE16_