Search

Boken Note To Self rymmer ett urval och en överblick av bildkonstnären Fredrik Hofwanders verk från det senaste decenniet fram till idag. Bilderna går i dialog med nyskrivna texter av författaren och mångåriga vännen Mats Kempe. Boken gavs ut i samband med vernissagen till Fredrik Hofwanders utställning En förgylld evighet, med verk som alla har staden Malmö som tema.

PURPOSE: The purpose of this study was to assess the long-term clinical, subjective, and radiographic results of pyrocarbon hemiarthroplasty for proximal interphalangeal joint (PIPJ) arthritis at a single institution. METHODS: Patients treated with a pyrolytic carbon hemiarthroplasty between 2005 and 2015 were contacted for a clinical follow-up visit. Patients were assessed before surgery, one year after surgery, and again after a mean of 11 years (range: 6-16 years). Objective outcomes were assessed with grip strength, pinch strength, and range of motion (ROM). Subjective outcomes were assessed by the Disabilities of the Arm, Shoulder, and Hand score, Canadian Occupational Performance Measure (performance and satisfaction), and Visual Analog Scale pain scores at rest and during activity. Radiographic assessments were completed according to Sweets and Stern as modified by Wagner et al. RESULTS: A total of 68 fingers in 52 patients underwent PIPJ hemiarthroplasty. Thirty-six arthroplasties in 29 patients were available for the long-term follow-up, five patients had died, and the remaining cases were contacted by phone. Three cases were lost to follow-up. Preoperative diagnoses included 41 fingers with osteoarthritis or posttraumatic arthritis, and 27 fingers with inflammatory arthritis. Eight cases had undergone revision at the time of follow-up, and the 10-year implant survival was 72%. The revisions were performed after a mean of two years after surgery. Three patients had undergone soft-tissue procedures. Visual Analog Scale pain scores, Disabilities of the Arm, Shoulder, and Hand scores, and Canadian Occupational Performance Measure scores improved significantly compared with that before surgery. Grip strength and pinch grip remained unchanged. However, PIPJ ROM deteriorated significantly one year after surgery, when compared with that before surgery. CONCLUSIONS: Pyrocarbon hemiarthroplasty of the PIPJ has an acceptable long-term implant survival, and the signif

The decomposition of numbers when solving subtraction tasks is regarded as more powerful than counting-based strategies. Still, many students fail to solve subtraction tasks despite using decomposition. To shed light upon this issue, we take a variation theoretical perspective (Marton, 2015) seeing learning as a function of discerning critical aspects and their relations of the object of learning. In this paper, we focus on what number relations students see in a three-digit subtraction task, and how they see them. We analyzed interview data from 55 second-grade students who used decomposition strategies to solve 204 − 193 =. The variation theory of learning was used to analyze what number relations the students experienced and how they experienced them, aiming to explain why they made errors even though they used presumably powerful strategies in their problem-solving. The findings show that students who simultaneously experienced within-number relations and between-number relations when solving the task succeeded in solving it, whereas those who did not do this failed. These findings have importance for understanding what students need to discern in order to be able to solve subtraction tasks in a proficient way.

Föreliggande studie syftar till att undersöka hur ett urval enhetschefer på sjukhus inom Västra Götalandsregionen upplever ledningens kommunikation och vilka förutsättningarkommunikationsinsatsen ger enhetschefer att kommunicera med sina medarbetare. Studien undersöker kommunikation i en förändringsprocess under 2023 och utgår från organisationskommunikationens begrepp meningsskapande, meningsgivande samt transparent kommunikation. För att tillhandahålla kvalitativ kännedom genomfördes materialinsamlingen med hjälp av sju individuella semi-strukturerade intervjuer. Reflexiv tematisk analys används som analysmetod för att identifiera mönster, kategorier och teman från empirin. Resultatet belyser betydelsen av tvåvägskommunikation, exempelvis dialoger, i en hierarkiskt styrd organisation och hur det skapar möjligheter att kommunicera utökad meningsskapande och meningsgivande. Resultatet visar även huruvida ett tolkningsbaserat kommunikationsuppdrag lämpar sig inom Västra Götalandsregionens sjukhus. Analysens resultat påvisar att det faktum att media hinnerrapportera om förändringen innan enhetscheferna kan samtala med sina medarbetare är ett hinder och resulterar i avsaknaden av förtroendegivande kommunikation., This study aims to examine how a sample of unit managers at hospitals within Västra Götalandsregionen experience the organizational managements communication and what prerequisites to communication efforts the unit managers conceive to communicate with their employees. The study examines communication during a change process during 2023 and is based on organizational communications concepts sensemaking, sensegiving and transparent communication. In order to provide qualitative knowledge, the empirical material was collected through seven individual semi-structured interviews. Reflexive thematic analysis is used as a method to identify patterns, categories and themes from the empirical data. The results highlight the importance of two-way communication, such as dialogues, in a hierarchical organization and how it creates opportunities to communicate extended sensemaking and sensegiving. The result also reflects on whether an interpretation-based communication mission is suitable in the hospitals of Västra Götalandsregionen. The study’s results clarify that the fact that the media reporting about the change before unit managers can communicate with their employees is an obstacle and leads to lack of trustworthy communication

The bacterial cell-wall peptidoglycan is made of glycan strands crosslinked by short peptide stems. Crosslinks are catalyzed by DD-transpeptidases (4,3-crosslinks) and LD-transpeptidases (3,3-crosslinks). However, recent research on non-model species has revealed novel crosslink types, suggesting the existence of uncharacterized enzymes. Here, we identify an LD-transpeptidase, LDTGo, that generates 1,3-crosslinks in the acetic-acid bacterium Gluconobacter oxydans. LDTGo-like proteins are found in Alpha- and Betaproteobacteria lacking LD3,3-transpeptidases. In contrast with the strict specificity of typical LD- and DD-transpeptidases, LDTGo can use non-terminal amino acid moieties for crosslinking. A high-resolution crystal structure of LDTGo reveals unique features when compared to LD3,3-transpeptidases, including a proline-rich region that appears to limit substrate access, and a cavity accommodating both glycan chain and peptide stem from donor muropeptides. Finally, we show that DD-crosslink turnover is involved in supplying the necessary substrate for LD1,3-transpeptidation. This phenomenon underscores the interplay between distinct crosslinking mechanisms in maintaining cell wall integrity in G. oxydans.

Machine learning models often perform poorly under subpopulation shifts in the data distribution. Developing methods that allow machine learning models to better generalize to such shifts is crucial for safe deployment in real-world settings. In this paper, we develop a family of group-aware prior (GAP) distributions over neural network parameters that explicitly favor models that generalize well under subpopulation shifts. We design a simple group-aware prior that only requires access to a small set of data with group information and demonstrate that training with this prior yields state-of-the-art performance -- even when only retraining the final layer of a previously trained non-robust model. Group aware-priors are conceptually simple, complementary to existing approaches, such as attribute pseudo labeling and data reweighting, and open up promising new avenues for harnessing Bayesian inference to enable robustness to subpopulation shifts., Comment: Published in Proceedings of the 27th International Conference on Artificial Intelligence and Statistics (AISTATS 2024)

As AI model size grows, neural scaling laws have become a crucial tool to predict the improvements of large models when increasing capacity and the size of original (human or natural) training data. Yet, the widespread use of popular models means that the ecosystem of online data and text will co-evolve to progressively contain increased amounts of synthesized data. In this paper we ask: How will the scaling laws change in the inevitable regime where synthetic data makes its way into the training corpus? Will future models, still improve, or be doomed to degenerate up to total (model) collapse? We develop a theoretical framework of model collapse through the lens of scaling laws. We discover a wide range of decay phenomena, analyzing loss of scaling, shifted scaling with number of generations, the ''un-learning" of skills, and grokking when mixing human and synthesized data. Our theory is validated by large-scale experiments with a transformer on an arithmetic task and text generation using the large language model Llama2.

Rankings are ubiquitous across many applications, from search engines to hiring committees. In practice, many rankings are derived from the output of predictors. However, when predictors trained for classification tasks have intrinsic uncertainty, it is not obvious how this uncertainty should be represented in the derived rankings. Our work considers ranking functions: maps from individual predictions for a classification task to distributions over rankings. We focus on two aspects of ranking functions: stability to perturbations in predictions and fairness towards both individuals and subgroups. Not only is stability an important requirement for its own sake, but -- as we show -- it composes harmoniously with individual fairness in the sense of Dwork et al. (2012). While deterministic ranking functions cannot be stable aside from trivial scenarios, we show that the recently proposed uncertainty aware (UA) ranking functions of Singh et al. (2021) are stable. Our main result is that UA rankings also achieve multigroup fairness through successful composition with multiaccurate or multicalibrated predictors. Our work demonstrates that UA rankings naturally interpolate between group and individual level fairness guarantees, while simultaneously satisfying stability guarantees important whenever machine-learned predictions are used.

In the era of proliferation of large language and image generation models, the phenomenon of "model collapse" refers to the situation whereby as a model is trained recursively on data generated from previous generations of itself over time, its performance degrades until the model eventually becomes completely useless, i.e the model collapses. In this work, we study this phenomenon in the setting of high-dimensional regression and obtain analytic formulae which quantitatively outline this phenomenon in a broad range of regimes. In the special case of polynomial decaying spectral and source conditions, we obtain modified scaling laws which exhibit new crossover phenomena from fast to slow rates. We also propose a simple strategy based on adaptive regularization to mitigate model collapse. Our theoretical results are validated with experiments.

The second-order properties of the training loss have a massive impact on the optimization dynamics of deep learning models. Fort & Scherlis (2019) discovered that a large excess of positive curvature and local convexity of the loss Hessian is associated with highly trainable initial points located in a region coined the "Goldilocks zone". Only a handful of subsequent studies touched upon this relationship, so it remains largely unexplained. In this paper, we present a rigorous and comprehensive analysis of the Goldilocks zone for homogeneous neural networks. In particular, we derive the fundamental condition resulting in excess of positive curvature of the loss, explaining and refining its conventionally accepted connection to the initialization norm. Further, we relate the excess of positive curvature to model confidence, low initial loss, and a previously unknown type of vanishing cross-entropy loss gradient. To understand the importance of excessive positive curvature for trainability of deep networks, we optimize fully-connected and convolutional architectures outside the Goldilocks zone and analyze the emergent behaviors. We find that strong model performance is not perfectly aligned with the Goldilocks zone, calling for further research into this relationship.

Chain-of-Thought (CoT) reasoning is known to improve Large Language Models both empirically and in terms of theoretical approximation power. However, our understanding of the inner workings and conditions of apparition of CoT capabilities remains limited. This paper helps fill this gap by demonstrating how CoT reasoning emerges in transformers in a controlled and interpretable setting. In particular, we observe the appearance of a specialized attention mechanism dedicated to iterative reasoning, which we coined "iteration heads". We track both the emergence and the precise working of these iteration heads down to the attention level, and measure the transferability of the CoT skills to which they give rise between tasks.

In the era of exceptionally data-hungry models, careful selection of the training data is essential to mitigate the extensive costs of deep learning. Data pruning offers a solution by removing redundant or uninformative samples from the dataset, which yields faster convergence and improved neural scaling laws. However, little is known about its impact on classification bias of the trained models. We conduct the first systematic study of this effect and reveal that existing data pruning algorithms can produce highly biased classifiers. At the same time, we argue that random data pruning with appropriate class ratios has potential to improve the worst-class performance. We propose a "fairness-aware" approach to pruning and empirically demonstrate its performance on standard computer vision benchmarks. In sharp contrast to existing algorithms, our proposed method continues improving robustness at a tolerable drop of average performance as we prune more from the datasets. We present theoretical analysis of the classification risk in a mixture of Gaussians to further motivate our algorithm and support our findings.

Adversarial examples have been shown to cause neural networks to fail on a wide range of vision and language tasks, but recent work has claimed that Bayesian neural networks (BNNs) are inherently robust to adversarial perturbations. In this work, we examine this claim. To study the adversarial robustness of BNNs, we investigate whether it is possible to successfully break state-of-the-art BNN inference methods and prediction pipelines using even relatively unsophisticated attacks for three tasks: (1) label prediction under the posterior predictive mean, (2) adversarial example detection with Bayesian predictive uncertainty, and (3) semantic shift detection. We find that BNNs trained with state-of-the-art approximate inference methods, and even BNNs trained with Hamiltonian Monte Carlo, are highly susceptible to adversarial attacks. We also identify various conceptual and experimental errors in previous works that claimed inherent adversarial robustness of BNNs and conclusively demonstrate that BNNs and uncertainty-aware Bayesian prediction pipelines are not inherently robust against adversarial attacks.

Under 2021 tillkom det över 600 000 streamingabbonemang bara i Sverige. Det är tydligt att filmtittande är en stor del av svenskarnas vardag. Men vilka attribut avgör hur svenskar väljer att konsumera filmer? Denna studie syftar till att utreda hur ett filmomslag i genren komedi kan designas utifrÃ¥n faktorerna typografi och färg. Studien inleddes med en enkätundersökning för att fÃ¥ en djupare förstÃ¥else för vilka egenskaper inom typografi och färg som var särskiljande för olika genrer. Undersökningen visade att genrerna komedi, skräck och romantik skiljde sig Ã¥t avsevärt. Studiens resultat väckte ett intresse för att undersöka genren komedis specifika utseende. Förstudien inleddes sÃ¥ledes med ytterligare en enkätundersökning specificerad till komedi, denna gÃ¥ng riktad till unga vuxna i Ã¥lder 18-35 Ã¥r. Utöver enkätundersökningen bestod förstudien även av en omvärldsanalys för att fÃ¥ en uppfattning av hur redan befintliga filmomslag inom de tre genrerna var utformade. Förstudiens slutliga resultat visade att filmomslag i genren komedi generellt använde sig av mycket mer färg än genrerna skräck och romantik samt rundade sans-seriffer. Insikterna frÃ¥n förstudien ledde fram till ett designarbete som utgick frÃ¥n relevanta teorier och som sedan utvärderades i tvÃ¥ steg med hjälp av semistrukturerade intervjuer. Deltagarna i de semistrukturerade intervjuerna fick ta del av framtaget material inom genren komedi och svara pÃ¥ frÃ¥gor gällande typografi och färg. Slutligen skapades ett filmomslag som uppfyllde syftet och ansÃ¥gs tillhöra genren komedi av mÃ¥lgruppen. Resultatet blev ett fiktivt filmomslag med mönstrad bakgrund i färgerna rosa och grön tillsammans med en illustrerad figur som skriker ut: The future awaits i orangea bokstäver. De attribut som hos mÃ¥lgruppen uppfattades tillhöra genren komedi var ett filmomslag med färgstarka färger och ett karaktäristiskt serietypsnitt., Examensarbetet är utfört vid Institutionen för teknik och naturvetenskap (ITN) vid Tekniska fakulteten, Linköpings universitet

Denna rapport behandlar produktutvecklingsarbetet av en portabel trådlös högtalare med solcellstekniken Powerfoyle. Projektet är ett kandidatexamensarbete inom Teknisk Design på KTH med samarbetspartnerna Urbanista och Exeger. Projektbeskrivningen har varit att undersöka möjligheten att använda Exegers solcellsteknik Powerfoyle i en portabel trådlös högtalare för Urbanista, och att sedan ge ett konceptuellt förslag på hur en sådan högtalare skulle se ut. Processen inleddes med att genomföra en grundlig marknadsanalys från vilken många kravspecifika antaganden har gjorts, följt av en genomgående teknisk analys av en portabel trådlös högtalare. Utifrån användarscenarion har integrationen av Powerfoyle i en trådlös högtalare simulerats. Slutsatsen är att om högtalaren får stå ca 2h per dag i ett soligt fönster över en vecka, så kan det kompensera för ca 16h speltid över en solig helg. Vidare i arbetet har ett slutgiltigt koncept utvecklats och fått namnet Urbanista Chile, en kilformad portabel högtalare, med Powerfoyle på baksidan. Högtalaren är klädd i en nylonmesh, över- och undersida är bestyckade med gummidetaljer. Chile är en högtalare som väcker uppmärksamhet och interaktion, den första i en ny kategori av högtalare., This report will cover the process of developing a portable speaker with the integrated solar cell technology Powerfoyle. The project is a bachelor's thesis in Industrial Design Engineering at the Royal Technical Institute, KTH, in Stockholm together with audio company Urbanista and deep tech company Exeger. The project description has been to explore the credibility in using Powerfoyle from Exeger in a wireless portable speaker for Urbanista, and furthermore to give a concept proposition of such a speaker The process includes a thorough market analysis from which many assumptions have been made, followed by an in-depth technical analysis of the components in a speaker. The integration of Powerfoyle and its performance in a portable speaker has gone through a simulation based on user scenarios. The conclusion is that if the speaker sits in a sunny window at least 2h a day for a week, it will sum up to a battery gain equal to the drain of playing music for 16h on a sunny weekend. The project finalizes in a conceptual speaker named Urbanista Chile, a wedge-shaped speaker with a Powerfoyle panel on the back. The product is wrapped in a nylon mesh with a top and bottom made from rubber. Chile is a speaker demanding attention and proposes interaction, it will be a pioneer in a new category of speakers.

We investigate a 2,000-year genetic transect through Scandinavia spanning the Iron Age to the present, based on 48 new and 249 published ancient genomes and genotypes from 16,638 modern individuals. We find regional variation in the timing and magnitude of gene flow from three sources: the eastern Baltic, the British-Irish Isles, and southern Europe. British-Irish ancestry was widespread in Scandinavia from the Viking period, whereas eastern Baltic ancestry is more localized to Gotland and central Sweden. In some regions, a drop in current levels of external ancestry suggests that ancient immigrants contributed proportionately less to the modern Scandinavian gene pool than indicated by the ancestry of genomes from the Viking and Medieval periods. Finally, we show that a north-south genetic cline that characterizes modern Scandinavians is mainly due to the differential levels of Uralic ancestry and that this cline existed in the Viking Age and possibly earlier.

The changes that have taken place during and after the Covid-19 pandemic are a fact, including the way managers communicate. This is because more and more employees are choosing to work from home, which represents a distance of communication between the manager and employee which they did not previously have. In contrast, remote communication in the consulting industry has been established since before the pandemic, where consulting managers are not present at the same workplace as the consultants. The aim of this study is to investigate the importance of managerial presence in communication between the manager and the employee in the consulting and staffing industry. To answer the aim, a comparison of the communication between a present and an absent manager is carried out. This is done using the concepts: communication, demands, control, support, availability and belonging. To answer the purpose, a comparative qualitative study has been conducted with three consultants and their respective consulting managers, where a comparison between the consultants' current workplace as a consultant and their then workplace when they were not working as a consultant is carried out. The empirical data is analyzed based on the theory of the requirement-controlsupport model, the theory of interaction and organizational identity. The study shows that clear communication and support in the form of the consultant manager's availability to the consultant contributes to an increased sense of control over the work and belonging to the consulting firm for the consultant. Thus, the fact that they do not communicate face-to-face does not matter, if the communication is clear, and the consultant feels a sense of security towards the consultant manager., Att omställningar skett under och efter Covid-19 pandemin är ett faktum, varav chefers sätt att kommunicera har varit en omställning. Orsaken till omställningen är att allt fler anställda väljer att arbeta hemifrån, vilket skapar en distanskommunikation mellan chef och den anställda som tidigare in existerat. Däremot är distanskommunikationen inom konsultbranschen etablerat sedan innan pandemin, där konsultchefer inte är närvarande på konsulternas arbetsplats. Syftet med studien är att undersöka vilken betydelse chefens närvaro har i kommunikationen mellan chefen och den anställda inom konsult- och bemanningsbranschen. För att besvara syftet har en jämförelse av kommunikationen mellan en närvarande respektive frånvarande chef genomförts med hjälp av begreppen: kommunikation, krav, kontroll, stöd, tillgänglighet samt tillhörighet. För att kunna besvara syftet har en komparativ kvalitativ studie genomförts med tre konsulter och deras respektive konsultchefer, där konsulternas tidigare arbetsplats, när de inte arbetade som konsult, och konsulternas nuvarande arbetsplats jämfördes. Empirin har analyserats utifrån teorin om krav-kontroll-stödmodellen, teori om interaktion samt organisationsidentitet. Studien visar att tydlig kommunikation och stöd i form av konsultchefens tillgänglighet bidrar till ökad känsla av kontroll över arbetet och tillhörighet till konsultföretaget för konsulten. Att de inte kommunicerar ansikte mot ansikte har således ingen betydelse, så länge kommunikationen är tydlig och konsulten känner en trygghet till sin konsultchef.

Dagens samhälle ställer allt högre krav på muntlig förmåga, i synnerhet i arbetslivet där allt fler yrken kräver hög muntlig kompetens. I takt med att kraven ökar blir talängslan ett allt större problem som leder till att talängsliga personer känner starka olustkänslor och därför undviker att tala inför andra människor. I gymnasieskolan är muntlig framställning ett obligatoriskt och betygsgrundande moment som kan vara svårt för talängsliga elever att både genomföra och lyckas med. Trots att talängslan är ett vanligt fenomen så är många lärare inte medvetna om hur de ska hantera talängsliga elever. Syftet med denna forskningsöversikt är därför att kartlägga hur talängslan i samspel med andra faktorer i lärmiljön påverkar elevers förberedda muntliga framställningar. Frågeställningen som forskningsöversikten ämnar besvara är: Vilka faktorer påverkar och har betydelse för talängsliga elevers utveckling av muntlig framställning i gymnasieskolans svenskundervisning? En systematisk sökprocess utmynnade i tio vetenskapliga artiklar med relevans för frågeställningen. Genom tematisk analysmetod identifierades fem teman som belyser betydande aspekter för talängsliga elevers muntliga utveckling. Vår slutsats är att nervositet och självförtroende, förberedelser, tidigare erfarenheter, övning inför muntlig framställning, stöttande lärmiljö och respons påverkar och har betydelse för talängsliga elevers utveckling av muntlig framställning. Den slutsats vi drar utifrån detta resultat kopplat till lärarprofessionen är att läraren framför allt har två uppdrag, dels att verka för en stöttande klassrumsmiljö, eftersom läraren har visat sig vara viktig vad gäller att skapa och upprätthålla en stöttande lärmiljö, dels att stötta enskilda elever med individuella anpassningar. Läraren behöver dessutom ha goda kunskaper kring muntlig framställning för att kunna tillämpa ett metaspråk vid respons som kan stötta och utveckla eleven på bästa sätt.

The independence of Scotland from Great Britain has been a long-standing and complex issue, with roots dating back centuries. In recent years, the question of independence has been brought to the forefront again with the UK’s decision to leave the European Union, commonly known as Brexit. After referendum in 2014, Scotland voted to remain part of the United Kingdom, but most Scottish voters chose to remain in the EU in the Brexit referendum of 2016. This has led to renewed calls for independence from some quarters, as many in Scotland feel that their interests are not being represented in the UK’s decision-making process. The issue of independence is highly contentious, with arguments for and against independence being made on economic, cultural and political grounds. Those in favor of independence argue that it would allow Scotland to have more control over its own affairs and to chart its own course in the world, while those opposed to independence argue that it would create unnecessary division and uncertainty, Ultimately the question of independence for Scotland remains a matter of ongoing debate on the will of the Scottish people and the decision of the UK government.

Denna studie har som mål att undersöka motivationsteorin grit och i vilken utsträckning dengår att relatera till akademiska resultat inom gymnasieskolan. Grit som konstrukt handlar omindividers passion och uthållighet för att nå långsiktiga mål. Undersökningen äger rum på enmindre gymnasieskola i södra Sverige där 23 elever från två olika grupper fick svara på enenkät utformad för att mäta grit. Snittbetyg, frånvaro och tidigare betyg inom kursernaMatematik 1b och Matematik 2b samlades in. Data analyserades därefter med hjälp av entematisk analys. Resultatet visade på en positiv svag korrelation mellan grit och snittbetygoch en svag negativ korrelation mellan grit och frånvaro. I regel uppvisade elever med höggrit något högre snittbetyg och lägre frånvaro. Spridningen bland individerna i studien varstora vilket är anledningen till att korrelationen uppvisade svaga samband. Det gick inte attuppvisa en korrelation utifrån betyg i matematik och grit. Trots svaga samband tyder studientillsammans med tidigare forskning på grit, eller motsvarande motivationsteorier, att det finnspositiva effekter av en implementation av motivationsgrundande teorier inom ramen förbefintlig undervisning.

We present a spatiotemporal picture of human genetic diversity in Anatolia, Iran, Levant, South Caucasus, and the Aegean, a broad region that experienced the earliest Neolithic transition and the emergence of complex hierarchical societies. Combining 35 new ancient shotgun genomes with 382 ancient and 23 present-day published genomes, we found that genetic diversity within each region steadily increased through the Holocene. We further observed that the inferred sources of gene flow shifted in time. In the first half of the Holocene, Southwest Asian and the East Mediterranean populations homogenized among themselves. Starting with the Bronze Age, however, regional populations diverged from each other, most likely driven by gene flow from external sources, which we term “the expanding mobility model.” Interestingly, this increase in inter-regional divergence can be captured by outgroup-f3-based genetic distances, but not by the commonly used FST statistic, due to the sensitivity of FST, but not outgroup-f3, to within-population diversity. Finally, we report a temporal trend of increasing male bias in admixture events through the Holocene.

Studiens syfte är att undersöka hur förskollärare beskriver barns delaktighet och inflytande på förskolan samt hur de anser sig kunna skapa förutsättningar för barns delaktighet och inflytande i förskolans utbildning. Studien utgår ifrån semistrukturerade intervjuer som metod där 10 förskollärare medverkat. Den teoretiska utgångspunkten för studien är det barndomssociologiska perspektivet med begreppen being och becoming, samt begreppet barns aktörskap och Shiers delaktighetsmodell. Resultatet synliggör förskollärares beskrivningar av strategier för att öka barns delaktighet och inflytande genom kartläggning av barns behov och intressen, tydliggörande pedagogik och demokratiska röstningar, men lyfter också hinder och utmaningar i förskolans uppdrag att främja barns delaktighet och inflytande. Studiens slutsats visar att förskollärare har en vilja att skapa förutsättningar för barns delaktighet och inflytande i förskolan, men att det finns osäkerhet i hur förskollärare kan öka barns delaktighet och inflytande i samspel med övriga mål och riktlinjer i förskolans styrdokument.

Detta är en kvalitativ studie med syftet att undersöka lärares erfarenheter och inställning till att kartlägga elevernas tidiga läs- och skrivutveckling och vilket/vilka kartläggningsmaterial som används. Studien genomfördes med sex semistrukturerade intervjuer med lärare från två olika skolor i två olika kommuner, fyra enskilda intervjuer och två parintervjuer. De två kartläggningsmaterial som främst användes är Nationellt bedömningsstöd i läs- och skrivutveckling i årskurs 1–3 och Legilexi, vilket beror på att kommunerna som lärarna är verksamma i har olika screeningplaner. Resultatet visar att lärarna har en positiv syn på kartläggning för att det kan ge läraren en tydlig överblick över vilken kunskap som redan finns i elevgruppen, både på grupp- och individnivå. Kartläggning är tidskrävande, men översikten som kartläggningen resulterar i stöttar läraren i sin planering av undervisningen för att kunna införa stöd och anpassningar där det behövs. Erfarenhet från mellanstadiet upplevdes fördelaktigt eftersom lärarna vet vad som komma skall för eleverna, på så sätt kan läraren stötta, pusha och förbereda eleverna inför mellanstadiet. För att kartlägga elevernas skrivutveckling använder lärarna egna skrivövningar och skrivuppgifter, utöver de områdena som Nationellt bedömningsstöd i läs- och skrivutveckling i årskurs 1–3 och Legilexi testar av inom skrivutvecklingen.

In this thesis, I delve into how Rhetorical Genre Study and Systemic-Functional Grammar can be used to assess the extent to which GPT-4 adheres to generic features and can be deemed adapted for its function. The objective is to establish a systematic model for objectively evaluating the degree to which AI-generated text is suitable for its intended purpose. To achieve this, I perform a rhetorical genre analysis on a crisis communication genre, which I subsequently quantify. I utilize the concept of topoi to pinpoint the arguments that serve to fulfill the functions of crisis communication. Subsequently, I prompt GPT-4 to produce texts within the genre and contrast them with my discoveries. Alongside this, I investigate the disparate results produced when using text and meta-text as input. The findings reveal a functional model to appraise the degree of text adaptation for its purpose within the analyzed genre. Although the model necessitates further fine-tuning, it effectively distinguishes nongeneric texts. The scope of the material selected was too limited to indicate any disparities in outcomes between different types of input. However, all input models generated texts that substantially conformed to the generic features.

Radical trapping agents such as Ferrostatin-1 (Fer-1) are capable of rescuing cells from ferroptosis, an iron-dependent form of cell death. Previously, poly(2-oxazoline)-Fer-1 (POx-Fer-1) conjugates were reported, which possess increased water-solubility and remain active after covalent conjugation of Fer-1. In this study, we break down the structural and functional layers of POx-Fer-1 conjugates and reveal that drug-free POx containing arylalkylamine and benzamide motifs show anti-ferroptosis properties. Intriguingly, even the basic construct poly(2-methyl-2-oxazoline-grad-2-phenyl-2-oxazoline) P(MeOx-grad-PhOx) was found to be active. Therefore, P(MeOx-grad-PhOx) of varying compositions were prepared, characterized by 1H NMR spectroscopy and size exclusion chromatography and investigated with regard to their self-assembly in aqueous solution and activity in an in vitro ferroptosis model. These findings were further explored for the design of defined and bioactive core-crosslinked micelles with intrinsic anti-ferroptosis behaviour. Cellular interaction studies involving C11-BODIPY assays and confocal microscopy investigations revealed lysosomal processing of the nanomaterials and perturbation of ferroptotic cell death through reducing lipid-peroxidation. This study highlights new drug/cargo-free anti-ferroptotic nanomaterials as proof of concept that hold potential for therapy of ferroptosis-associated diseases and highlights the role of nanocarriers in a therapeutic context.

Bakgrund: År 2022 kantades av Rysslands invasion av Ukraina, börsnedgång, stigande inflation och styrräntor. Det bidrog till att allt fler sökte hjälp av finansiella rådgivare för att utifrån livssituationen få hjälp med att nå sina sparmål och maximera sin avkastning. I tidigare forskning råder det delade meningar kring om den finansiella rådgivaren faktiskt bidrar till positiva ekonomiska utfall hos kunden. Å andra sidan får den finansiella rådgivaren, åtminstone privat, kunskap till sig via sitt arbete som torde leda till en hög finansiell bildning. Huruvida det faktiskt bidrar till positiva ekonomiska utfall i den finansiella rådgivarens egna investeringsbeslut och sparande under 2022 leder således till frågan, om de kan hjälpa andra, kan de även hjälpa sig själva? Syfte: Studien syftar till att skapa en förståelse för hur finansiella rådgivares privata sparande och investeringsbeslut samvarierar med olika faktorer såsom livssituation, psykologiska bias och erfarenheter under 2022. Vidare ämnar studien utforska hur finansiella rådgivare ställer sig till ytterligare utbildning inom beteendeekonomi och huruvida de tror att det kan bidra till bättre finansiellt beslutsfattande. Metod: En kvalitativ metod har använts för att uppfylla studiens syfte. Tio finansiella rådgivare intervjuades genom semistrukturerade djupintervjuer. Slutsats: Studien finner att placering i livscykeln, psykologiska bias samt erfarenhet samvarierar med den finansiella rådgivarens privata investeringsbeslut och sparande under 2022. Placeringen i livscykeln samvarierar främst med riskaversionen, därutöver har de finansiella rådgivarna inte kunnat motstå fall av samtliga undersökta bias i studien. Erfarenheten samvarierar med en minskad benägenhet att falla offer för vissa psykologiska bias, förutom överkonfidens, där effekten snarare är den motsatta. Vidare anser de finansiella rådgivarna att ytterligare utbildning inom beteendeekonomi inte skulle vara värdeskapande för deras privata invester, Background: The year 2022 was characterized by the invasion of Ukraine, the downturn of the stock market, rising inflation and increases in the benchmark interest rate. This led to the increasing search for financial advisors who, using life circumstances, could assist the individual in reaching personal saving goals and maximizing returns on investments. In prior research, there are divided opinions about whether financial advisors contribute to positive financial outcomes for the client. The financial advisors, at least privately, gain knowledge through their work which should lead to a high level of financial literacy. Whether it contributes to positive financial outcomes in the financial advisor's own investment decisions and savings in 2022 thus leads to the question, if they can help others, can they also help themselves? Aim: The study aims to create an understanding of how financial advisers' private investments decisions and savings in 2022 have been affected by various factors such as life situation, psychological bias, and experiences. Furthermore, the study intends to explore financial advisers' perception of how further education in behavioral economics could have contributed to better financial decision-making. Method: A qualitative method has been used to fulfill the purpose of the study. Ten financial advisors were interviewed through semi-structured in-depth interviews. Conclusion: The study finds that placement in the life cycle, psychological bias, and experience had an impact on financial advisors’ private investment decisions and savings during 2022. The placement in the lifecycle primarily influences the risk aversion, while the financial advisors have not been able to withstand cases of all investigated biases in the study. Experience has contributed to mitigating the impact of psychological biases to some extent while also creating overconfidence. The financial advisors believe that additional education in behavioral economics would not add v

Syftet med studien var att skapa förståelse för förskollärares uppfattningar av det kompensatoriska uppdraget i relation till teknikundervisningens såväl som deras förhållningssätt till teknik och teknikämnet i sig. Vidare syftar arbetet till att åskådliggöra hur barns förhållningssätt till teknik kommer till uttryck i undervisningssituationer. För att samla in empiri till studien använde vi oss av ljudinspelning, anteckningar och observationer tillsammans med fem förskollärare och deras respektive barngrupp. Genom dessa metodval kunde vi därmed synliggöra barnens och förskollärarnas erfarenheter av och sätt att framställa teknik. Vidare använde vi oss därefter av semistrukturerade intervjuer och även där användes ljudinspelning. Detta för att ytterligare synliggöra hur förskollärarna framställer det kompensatoriska uppdraget i relation till teknikinnehållet i undervisningen.Det inspelade materialet från observationerna och intervjuerna transkriberades och analyserades utifrån en innehållsanalys och studiens teoretiska utgångspunkter; sociokulturella teorin samt en del av variationsteorin. Inledningsvis visar resultatet att förskollärarna uppfattar teknik på olika sätt. Dessa olika sätt var teknik som problemlösning i samband med de mäktiga fem vilket bland annat består av det lutande plan, hjulet, kilen, hävstången och skruven. Teknik har även uppfattas i relation till fysik samt tekniska artefakter. Resultatet visar även att barnen uppfattar tekniska artefakter med hjälp av förskollärarnas framställning av de mäktiga fem. Barnen sätter de tekniska artefakterna i relation till deras vardag. Samspel och interaktion med andra och de tekniska artefakterna i barnens vardag anses vara betydelsefullt för deras utveckling och lärande. Resultatet visar att det kompensatoriska uppdraget innefattar flera viktiga aspekter för att barnen ska ges möjlighet att erfara och skapa förståelse för vad teknik kan ingripa, både inom och utanför förskolans kontext. Dessa aspekter är del

Syftet med denna artikel är att beskriva och diskutera vilka, för lärare, användbara ochrelevanta kunskapsprodukter som har genererats i 11 forskningsprojekt, finansieradeav Skolforskningsinstitutet och beviljade år 2016 och 2017. Forskningsfrågorna är: Vad iresultaten kan ses som kunskapsprodukter och vilken karaktär har dessa? På vilka sätt harlärare medverkat i processen att ta fram dessa kunskapsprodukter? Sammantaget har 55öppet tillgängliga granskade publikationer analyserats. Sex kvalitativt skilda kategorierav kunskapsprodukter har identifierats: 1) Kunnanden och förmågor, 2) Undervisnings- ochlektionsdesign, 3) Didaktiska exempel, 4) Redskap, 5) Processer och metaperspektiv och 6)Generella eller kontextuella förutsättningar. Graden av medverkan från lärares sida har varieratavsevärt; från en aktiv medverkan, där lärare och forskare samarbetar på likvärdigavillkor, till en låg grad, där lärare primärt fungerar som bollplank eller forskningsobjekt., This article examines the knowledge products of relevance to the teaching professionfrom eleven research projects funded by the Swedish Institute for Educational Research.A total of 55 openly accessible reviewed publications have been analysed. The aim is todescribe and discuss which, for teachers, useful and relevant knowledge products thatthe research projects (granted 2016 and 2017) have yielded. The research questions are:What in the results can be interpreted as knowledge products and what is the characteristicof these? In what ways have teachers participated in the process of producingthese knowledge products? Six qualitatively different categories of knowledge productshave been identified as descriptions of: 1) Knowings and capabilities, 2) Teaching/lessondesign, 3) Didactical examples, 4) Tools, 5) Processes and meta-perspectives, and 6)General or contextual preconditions. The degree of involvement of teachers has variedconsiderably, ranging from a high degree of active involvement, where teachers andresearchers collaborate on equal terms, to a low degree of involvement where teachersprimarily act as informants.

BACKGROUND: Training load is typically described in terms of internal and external load. Investigating the coupling of internal and external training load is relevant to many sports. Here, continuous kernel-density estimation (KDE) may be a valuable tool to capture and visualize this coupling. AIM: Using training load data in speed skating, we evaluated how well bivariate KDE plots describe the coupling of internal and external load and differentiate between specific training sessions, compared to training impulse scores or intensity distribution into training zones. METHODS: On-ice training sessions of 18 young (sub)elite speed skaters were monitored for velocity and heart rate during 2 consecutive seasons. Training session types were obtained from the coach's training scheme, including endurance, interval, tempo, and sprint sessions. Differences in training load between session types were assessed using Kruskal-Wallis or Kolmogorov-Smirnov tests for training impulse and KDE scores, respectively. RESULTS: Training impulse scores were not different between training session types, except for extensive endurance sessions. However, all training session types differed when comparing KDEs for heart rate and velocity (both P < .001). In addition, 2D KDE plots of heart rate and velocity provide detailed insights into the (subtle differences in) coupling of internal and external training load that could not be obtained by 2D plots using training zones. CONCLUSION: 2D KDE plots provide a valuable tool to visualize and inform coaches on the (subtle differences in) coupling of internal and external training load for training sessions. This will help coaches design better training schemes aiming at desired training adaptations.

Metabolic oligosaccharide engineering (MOE) of cells with synthetic monosaccharides can introduce functionality to the glycans of cell membranes. Unnatural sugars (e. g., peracetylated mannose-azide) can be expressed on the cell surface with the azide group in place. After MOE, the azide group can participate in a copper-free click reaction with an alkyne (e. g., dibenzocyclooctyne, DBCO) probe. This allows the metabolic fate of monosaccharides in cells to be understood. However, in a drug delivery context it is desirable to have azide groups on the probe (e. g. a drug delivery particle) and the alkyne (e. g. DBCO) on the cell surface. Consequently, the labelling efficiency of intestinal cell lines (Caco-2 and HT29-MTX-E12) treated with N-dibenzocyclooctyne-tetra-acetylmannosamine, and the concentration- and time-dependent labelling were determined. Furthermore, the labelling of mucus in HT29-MTX-E12 cells with DBCO was shown. This study highlights the potential for using MOE to target azide-functionalised probes to intestinal tissues for drug delivery applications.

Objectives: Evaluating effectiveness of speech/phrase recognition software in critically ill patients with speech impairments. Design: Prospective study. Setting: Tertiary hospital critical care unit in the northwest of England. Participants: 14 patients with tracheostomies, 3 female and 11 male. Main outcome measures: Evaluation of dynamic time warping (DTW) and deep neural networks (DNN) methods in a speech/phrase recognition application. Using speech/phrase recognition app for voice impaired (SRAVI), patients attempted mouthing various supported phrases with recordings evaluated by both DNN and DTW processing methods. Then, a trio of potential recognition phrases was displayed on the screen, ranked from first to third in order of likelihood. Results: A total of 616 patient recordings were taken with 516 phrase identifiable recordings. The overall results revealed a total recognition accuracy across all three ranks of 86% using the DNN method. The rank 1 recognition accuracy of the DNN method was 75%. The DTW method had a total recognition accuracy of 74%, with a rank 1 accuracy of 48%. Conclusion: This feasibility evaluation of a novel speech/phrase recognition app using SRAVI demonstrated a good correlation between spoken phrases and app recognition. This suggests that speech/phrase recognition technology could be a therapeutic option to bridge the gap in communication in critically ill patients. What is already known about this topic: Communication can be attempted using visual charts, eye gaze boards, alphabet boards, speech/phrase reading, gestures and speaking valves in critically ill patients with speech impairments. What this study adds: Deep neural networks and dynamic time warping methods can be used to analyse lip movements and identify intended phrases. How this study might affect research, practice and policy: Our study shows that speech/phrase recognition software has a role to play in bridging the communication gap in speech impairment.

We introduce a search problem generalizing the typical setting of Binary Search on the line. Similar to the setting for Binary Search, a target is chosen adversarially on the line, and in response to a query, the algorithm learns whether the query was correct, too high, or too low. Different from the Binary Search setting, the cost of a query is a monotone non-decreasing function of the distance between the query and the correct answer; different functions can be used for queries that are too high vs. those that are too low. The algorithm's goal is to identify an adversarially chosen target with minimum total cost. Note that the algorithm does not even know the cost it incurred until the end, when the target is revealed. This abstraction captures many natural settings in which a principal experiments by setting a quantity (such as an item price, bandwidth, tax rate, medicine dosage, etc.) where the cost or regret increases the further the chosen setting is from the optimal one. First, we show that for arbitrary symmetric cost functions (i.e., overshooting vs. undershooting by the same amount leads to the same cost), the standard Binary Search algorithm is a 4-approximation. We then show that when the cost functions are bounded-degree polynomials of the distance, the problem can be solved optimally using Dynamic Programming; this relies on a careful encoding of the combined cost of past queries (which, recall, will only be revealed in the future). We then generalize the setting to finding a node on a tree; here, the response to a query is the direction on the tree in which the target is located, and the cost is increasing in the distance on the tree from the query to the target. Using the k-cut search tree framework of Berendsohn and Kozma and the ideas we developed for the case of the line, we give a PTAS when the cost function is a bounded-degree polynomial., Comment: 21 pages, 3 figures

The prevalence and importance of algorithmic two-sided marketplaces has drawn attention to the issue of fairness in such settings. Algorithmic decisions are used in assigning students to schools, users to advertisers, and applicants to job interviews. These decisions should heed the preferences of individuals, and simultaneously be fair with respect to their merits (synonymous with fit, future performance, or need). Merits conditioned on observable features are always \emph{uncertain}, a fact that is exacerbated by the widespread use of machine learning algorithms to infer merit from the observables. As our key contribution, we carefully axiomatize a notion of individual fairness in the two-sided marketplace setting which respects the uncertainty in the merits; indeed, it simultaneously recognizes uncertainty as the primary potential cause of unfairness and an approach to address it. We design a linear programming framework to find fair utility-maximizing distributions over allocations, and we show that the linear program is robust to perturbations in the estimated parameters of the uncertain merit distributions, a key property in combining the approach with machine learning techniques., Comment: 24 pages, 2 figures. New draft includes experiments and lower bound

Scandinavia spanning the Iron Age to the present, based on 48 new and 249 published ancient genomes and genotypes from 16,638 modern individuals. We find regional variation in the timing and magnitude of gene flow from three sources: the eastern Baltic, the British-Irish Isles, and southern Europe. British-Irish ancestry was widespread in Scandinavia from the Viking period, whereas eastern Baltic ancestry is more localized to Gotland and central Sweden. In some regions, a drop in current levels of external ancestry suggests that ancient immigrants contributed proportionately less to the modern Scandinavian gene pool than indicated by the ancestry of genomes from the Viking and Medieval periods. Finally, we show that a north-south genetic cline that characterizes modern Scandinavians is mainly due to the differential levels of Uralic ancestry and that this cline existed in the Viking Age and possibly earlier.

We revisit the recent breakthrough result of Gkatzelis et al. on (single-winner) metric voting, which showed that the optimal distortion of 3 can be achieved by a mechanism called Plurality Matching. The rule picks an arbitrary candidate for whom a certain candidate-specific bipartite graph contains a perfect matching, and thus, it is not neutral (i.e, symmetric with respect to candidates). Subsequently, a much simpler rule called Plurality Veto was shown to achieve distortion 3 as well. This rule only constructs such a matching implicitly but the winner depends on the order that voters are processed, and thus, it is not anonymous (i.e., symmetric with respect to voters). We provide an intuitive interpretation of this matching by generalizing the classical notion of the (proportional) veto core in social choice theory. This interpretation opens up a number of immediate consequences. Previous methods for electing a candidate from the veto core can be interpreted simply as matching algorithms. Different election methods realize different matchings, in turn leading to different sets of candidates as winners. For a broad generalization of the veto core, we show that the generalized veto core is equal to the set of candidates who can emerge as winners under a natural class of matching algorithms reminiscent of Serial Dictatorship. Extending these matching algorithms into continuous time, we obtain a highly practical voting rule with optimal distortion 3, which is also intuitive and easy to explain: Each candidate starts off with public support equal to his plurality score. From time 0 to 1, every voter continuously brings down, at rate 1, the support of her bottom choice among not-yet-eliminated candidates. A candidate is eliminated if he is opposed by a voter after his support reaches 0. On top of being anonymous and neutral, this rule satisfies many other axioms desirable in practice., Comment: 24 pages, 1 figure, EC'23

Research on improving the robustness of neural networks to adversarial noise - imperceptible malicious perturbations of the data - has received significant attention. The currently uncontested state-of-the-art defense to obtain robust deep neural networks is Adversarial Training (AT), but it consumes significantly more resources compared to standard training and trades off accuracy for robustness. An inspiring recent work [Dapello et al.] aims to bring neurobiological tools to the question: How can we develop Neural Nets that robustly generalize like human vision? [Dapello et al.] design a network structure with a neural hidden first layer that mimics the primate primary visual cortex (V1), followed by a back-end structure adapted from current CNN vision models. It seems to achieve non-trivial adversarial robustness on standard vision benchmarks when tested on small perturbations. Here we revisit this biologically inspired work, and ask whether a principled parameter-free representation with inspiration from physics is able to achieve the same goal. We discover that the wavelet scattering transform can replace the complex V1-cortex and simple uniform Gaussian noise can take the role of neural stochasticity, to achieve adversarial robustness. In extensive experiments on the CIFAR-10 benchmark with adaptive adversarial attacks we show that: 1) Robustness of VOneBlock architectures is relatively weak (though non-zero) when the strength of the adversarial attack radius is set to commonly used benchmarks. 2) Replacing the front-end VOneBlock by an off-the-shelf parameter-free Scatternet followed by simple uniform Gaussian noise can achieve much more substantial adversarial robustness without adversarial training. Our work shows how physically inspired structures yield new insights into robustness that were previously only thought possible by meticulously mimicking the human cortex., Comment: Machine Learning and the Physical Sciences Workshop, NeurIPS 2022

The conference peer review process involves three constituencies with different objectives: authors want their papers accepted at prestigious venues (and quickly), conferences want to present a program with many high-quality and few low-quality papers, and reviewers want to avoid being overburdened by reviews. These objectives are far from aligned, primarily because the evaluation of a submission is inherently noisy. Over the years, conferences have experimented with numerous policies to navigate the tradeoffs. These experiments include setting various bars for acceptance, varying the number of reviews per submission, requiring prior reviews to be included with resubmissions, and others. In this work, we investigate, both analytically and empirically, how well various policies work, and more importantly, why they do or do not work. We model the conference-author interactions as a Stackelberg game in which a prestigious conference commits to an acceptance policy; the authors best-respond by (re)submitting or not (re)submitting to the conference in each round of review, the alternative being a "sure accept" (such as a lightly refereed venue). Our main results include the following observations: 1) the conference should typically set a higher acceptance threshold than the actual desired quality; we call this the "resubmission gap". 2) the reviewing load is heavily driven by resubmissions of borderline papers - therefore, a judicious choice of acceptance threshold may lead to fewer reviews while incurring an acceptable loss in conference quality. 3) conference prestige, reviewer inaccuracy, and author patience increase the resubmission gap, and thus increase the review load for a fixed level of conference quality. For robustness, we further consider different models of paper quality and compare our theoretical results to simulations based on plausible parameters estimated from real data.

Background and aimsProblem gambling and tobacco use are highly comorbid among adults. However, there are few treatment frameworks that target both gambling and tobacco use simultaneously (i.e., an integrated approach), while also being accessible and evidence-based. The aim of this two-arm open label RCT was to examine the efficacy of an integrated online treatment for problem gambling and tobacco use.MethodsA sample of 209 participants (M$_{age}$ = 37.66, SD = 13.81; 62.2% female) from North America were randomized into one of two treatment conditions (integrated [n = 91] or gambling only [n = 118]) that lasted for eight weeks and consisted of seven online modules. Participants completed assessments at baseline, after treatment completion, and at 24-week follow-up.ResultsWhile a priori planned generalized linear mixed models showed no condition differences on primary (gambling days, money spent, time spent) and secondary outcomes, both conditions did appear to significantly reduce problem gambling and smoking behaviours over time. Post hoc analyses showed that reductions in smoking and gambling craving were correlated with reductions in days spent gambling, as well as with gambling disorder symptoms. Relatively high (versus low) nicotine replacement therapy use was associated with greater reductions in gambling behaviours in the integrated treatment condition.Discussion and conclusionsWhile our open label RCT does not support a clear benefit of integrated treatment, findings suggest that changes in smoking and gambling were correlated over time, regardless of treatment condition, suggesting that more research on mechanisms of smoking outcomes in the context of gambling treatment may be relevant.

In many applications, Neural Nets (NNs) have classification performance on par or even exceeding human capacity. Moreover, it is likely that NNs leverage underlying features that might differ from those humans perceive to classify. Can we "reverse-engineer" pertinent features to enhance our scientific understanding? Here, we apply this idea to the notoriously difficult task of galaxy classification: NNs have reached high performance for this task, but what does a neural net (NN) "see" when it classifies galaxies? Are there morphological features that the human eye might overlook that could help with the task and provide new insights? Can we visualize tracers of early evolution, or additionally incorporated spectral data? We present a novel way to summarize and visualize galaxy morphology through the lens of neural networks, leveraging Dataset Distillation, a recent deep-learning methodology with the primary objective to distill knowledge from a large dataset and condense it into a compact synthetic dataset, such that a model trained on this synthetic dataset achieves performance comparable to a model trained on the full dataset. We curate a class-balanced, medium-size high-confidence version of the Galaxy Zoo 2 dataset, and proceed with dataset distillation from our accurate NN-classifier to create synthesized prototypical images of galaxy morphological features, demonstrating its effectiveness. Of independent interest, we introduce a self-adaptive version of the state-of-the-art Matching Trajectory algorithm to automate the distillation process, and show enhanced performance on computer vision benchmarks., Comment: Accepted to NeurIPS Workshop on Machine Learning and the Physical Sciences, 2023

We introduce a novel definition for a small set R of k points being "representative" of a larger set in a metric space. Given a set V (e.g., documents or voters) to represent, and a set C of possible representatives, our criterion requires that for any subset S comprising a theta fraction of V, the average distance of S to their best theta*k points in R should not be more than a factor gamma compared to their average distance to the best theta*k points among all of C. This definition is a strengthening of proportional fairness and core fairness, but - different from those notions - requires that large cohesive clusters be represented proportionally to their size. Since there are instances for which - unless gamma is polynomially large - no solutions exist, we study this notion in a resource augmentation framework, implicitly stating the constraints for a set R of size k as though its size were only k/alpha, for alpha > 1. Furthermore, motivated by the application to elections, we mostly focus on the "ordinal" model, where the algorithm does not learn the actual distances; instead, it learns only for each point v in V and each candidate pairs c, c' which of c, c' is closer to v. Our main result is that the Expanding Approvals Rule (EAR) of Aziz and Lee is (alpha, gamma) representative with gamma <= 1 + 6.71 * (alpha)/(alpha-1). Our results lead to three notable byproducts. First, we show that the EAR achieves constant proportional fairness in the ordinal model, giving the first positive result on metric proportional fairness with ordinal information. Second, we show that for the core fairness objective, the EAR achieves the same asymptotic tradeoff between resource augmentation and approximation as the recent results of Li et al., which used full knowledge of the metric. Finally, our results imply a very simple single-winner voting rule with metric distortion at most 44., Comment: 24 pages, Accepted to AAAI 24

Neural Collapse refers to the curious phenomenon in the end of training of a neural network, where feature vectors and classification weights converge to a very simple geometrical arrangement (a simplex). While it has been observed empirically in various cases and has been theoretically motivated, its connection with crucial properties of neural networks, like their generalization and robustness, remains unclear. In this work, we study the stability properties of these simplices. We find that the simplex structure disappears under small adversarial attacks, and that perturbed examples "leap" between simplex vertices. We further analyze the geometry of networks that are optimized to be robust against adversarial perturbations of the input, and find that Neural Collapse is a pervasive phenomenon in these cases as well, with clean and perturbed representations forming aligned simplices, and giving rise to a robust simple nearest-neighbor classifier. By studying the propagation of the amount of collapse inside the network, we identify novel properties of both robust and non-robust machine learning models, and show that earlier, unlike later layers maintain reliable simplices on perturbed data.

Dataset distillation extracts a small set of synthetic training samples from a large dataset with the goal of achieving competitive performance on test data when trained on this sample. In this work, we tackle dataset distillation at its core by treating it directly as a bilevel optimization problem. Re-examining the foundational back-propagation through time method, we study the pronounced variance in the gradients, computational burden, and long-term dependencies. We introduce an improved method: Random Truncated Backpropagation Through Time (RaT-BPTT) to address them. RaT-BPTT incorporates a truncation coupled with a random window, effectively stabilizing the gradients and speeding up the optimization while covering long dependencies. This allows us to establish new state-of-the-art for a variety of standard dataset benchmarks. A deeper dive into the nature of distilled data unveils pronounced intercorrelation. In particular, subsets of distilled datasets tend to exhibit much worse performance than directly distilled smaller datasets of the same size. Leveraging RaT-BPTT, we devise a boosting mechanism that generates distilled datasets that contain subsets with near optimal performance across different data budgets., Comment: Short version appears at NeurIPS 2023 WANT workshop

Lecture notes from the course given by Professor Julia Kempe at the summer school "Statistical physics of Machine Learning" in Les Houches. The notes discuss the so-called NTK approach to problems in machine learning, which consists of gaining an understanding of generally unsolvable problems by finding a tractable kernel formulation. The notes are mainly focused on practical applications such as data distillation and adversarial robustness, examples of inductive bias are also discussed., Comment: These are notes from the lecture of Julia Kempe given at the summer school "Statistical Physics \& Machine Learning", that took place in Les Houches School of Physics in France from 4th to 29th July 2022

We explored use of Recurrence Quantification Analysis (RQA) of speech rhythm data from mental-health counseling sessions for prediction of quality of psychotherapy. Time-series of inter-syllable intervals (ISIs) were extracted from 239 counseling sessions conducted by 12 therapists who repeatedly interacted with 30 clients. We found a negative association between recurrence metrics and client-rated session quality and a negative link between percent of laminarity and therapist-rated session quality, after controlling for self-reported client depression and distress measures and duration of speech sound within a session. Placing value on reduced recurrence in patterns of ISIs, and especially reduced degree of a dyadic system remaining in the same speech-rhythm pattern may be indicative of a desire for variation in content and strategies of client-therapist interaction. These exploratory findings point to the possibility of RQA-based automated systems to capture the ‘footprint’ of the non-verbal dynamic that is indicative of successful mental-health counseling.