Objective To develop and validate a wrist‐worn accelerometer‐based, deep‐learning tunable algorithm for the automated detection of generalized or bilateral convulsive seizures (CSs) to be integrated with off‐the‐shelf smartwatches.

Methods We conducted a prospective multi‐center study across eight European epilepsy monitoring units, collecting data from 384 patients undergoing video electroencephalography (vEEG) monitoring with a wrist‐worn three dimensional (3D)–accelerometer sensor. We developed an ensemble‐based convolutional neural network architecture with tunable sensitivity through quantile‐based aggregation. The model, referred to as Episave, used accelerometer amplitude as input. It was trained on data from 37 patients who had 54 CSs and evaluated on an independent dataset comprising 347 patients, including 33 who had 49 CSs.

Results Cross‐validation on the training set showed that optimal performance was obtained with an aggregation quantile of 60, with a 98% sensitivity, and a false alarm rate (FAR) of 1/6 days. Using this quantile on the independent test set, the model achieved a 96% sensitivity (95% confidence interval [CI]: 90%–100%), a FAR of <1/8 days (95% CI: 1/9–1/7 days) with 1 FA/61 nights, and a median detection latency of 26 s. One of the two missed CSs could be explained by the patient's arm, which was wearing the sensor, being trapped in the bed rail. Other quantiles provided up to 100% sensitivity at the cost of a greater FAR (1/2 days) or very low FAR (1/100 days) at the cost of lower sensitivity (86%).

Significance This Phase 2 clinical validation study suggests that deep learning techniques applied to single‐sensor accelerometer data can achieve high CS detection performance while enabling tunable sensitivity.

Controlling the spin current lies at the heart of spintronics and its applications. In ferromagnets, the sign of spin currents is fixed once the current direction is determined. However, spin currents in antiferromagnets can possess opposite polarizations, but this requires enormous magnetic fields to lift the degeneracy between the two modes. Therefore, controlling spin currents with opposite polarization is still a challenge. Here we demonstrate the control of spin currents at room temperature by magnon interference in a canted antiferromagnet, namely, haematite that has recently been classified as an altermagnet. Magneto-optical characterization by Brillouin light scattering reveals that the spatial periodicity of the beating patterns is tunable via the microwave frequency. We further observe that the inverse spin Hall voltage changes sign as the frequency is tuned, evincing a frequency-controlled switching of polarization of pure spin currents. Our work highlights the use of antiferromagnetic magnon interference to control spin currents, which substantially extends the horizon for the emerging field of coherent antiferromagnetic spintronics.

Magnons are the quanta of spin waves and transport angular momenta through magnetically ordered materials. They can be used to distribute and control on-chip GHz signals without charge flow, thereby avoiding Joule heating. Beyond multiplexed signal processing, filtering, and Boolean logic, they allow for hardware implementation of neural networks exploiting cascaded magnon scattering on the nanoscale. A game-changing boost is expected if nonvolatile magnon-signal storage and in-memory computation schemes become realistic. We outline recent progress in experimental research and micromagnetic modeling toward these goals before sketching remaining challenges.

Après les splendeurs de la construction romaine, les compétences techniques et les connaissances relatives au béton ne se transmettent que localement et en sous-main, jusqu’à ce qu’au cours du XVIIIe siècle, dans plusieurs régions d’Europe, de nombreux chercheurs et constructeurs s’intéressent à l’étude de la chaux, des mortiers et des ciments. Les recherches sur les roches se multiplient afin de découvrir des matières premières facilement disponibles dans la nature pour produire des mortiers qui soient non seulement résistants, durs et compacts, mais aussi capables de durcir sous l’eau. Ces travaux sont stimulés tant par l’étude de l’archéologie et par la découverte de diverses oeuvres d’ingénierie hydraulique réalisées par les Romains, dont la résistance prend des connotations mythiques dans la littérature technique de l’époque, que par un besoin croissant de matériaux utiles à la construction de grandes oeuvres pour le développement du commerce (ports, ponts, canaux navigables et écluses) ou pour la défense militaire. À cette combinaison d’intérêts théoriques et pratiques s’ajoute la passion pour les explorations géographiques et les études naturalistes soutenues par les outils de la chimie moderne qui, depuis le siècle précédent, prend progressivement la place de l’alchimie, à la fois fascinante et insuffisamment vérifiable. Ainsi, des concepts au caractère presque magique, comme celui de phlogistique, sont abandonnés, tandis qu’à la moitié du XVIIIe siècle, le médecin et chimiste écossais Joseph Black découvre le cycle de transformation de la pierre à chaux en chaux, puis à nouveau en pierre à chaux. La découverte de l’origine de la capacité de certaines chaux à durcir dans l’eau, également appelée propriété hydraulique, est due aux recherches de nombreux chimistes et est attribuée, entre la fin du XVIIIe siècle et le début du XIXe siècle, à la présence de minéraux argileux dans certaines pierres calcaires, ouvrant ainsi la voie vers l’invention du ciment moderne, devenu célèbre sous le nom commercial de Portland.

Since 1989, and every 5 years, a group of database system researchers gather to reflect on the state of our field, to discuss the key directions the community is pursuing, and to speculate about the opportunities that lie ahead. The last meeting took place in Cambridge, Massachusetts, USA  in October 2023.

Settlement Freidorf near Basel was the first global cooperative in Switzerland, in which a group of people organized parts of their lives - such as housing, consumption, and education - collectively instead of privately. It was commissioned by the Union of Swiss Consumer Associations (USC) and completed in 1924 as the first large project of Swiss architect Hannes Meyer (1889-1954).

This thesis aims at a deeper and comprehensive understanding of this work of architecture, as a case study depicting the possibilities and limitations of co-operative housing in general. As a prologue, two historical sources are presented. On the one hand, there is the influential novel The Goldmakers' Village (Das Goldmacherdorf) from 1817, by German writer Heinrich Zschokke, considered as the first co-op novel. This work of fiction is a founding document of the ideas and theories that led to Freidorf. On the other hand, there is the black-and-white film made by Fritz MattmÃŒller between 1920 and 1923, showing the construction of Freidorf. These images are silent witnesses of the material genesis of the Siedlung, but also of the way it was used and inhabited following completion.

The actual thesis consists of five parts. The first part is an overview of the literature on both Freidorf and its architect, to show how both have been historized and interpreted so far. The second part is a minute description of Meyer's biography leading up to 1919, in which he first visited the building site in Muttenz. The third part is a reconstruction of the social history and the theoretical arguments, mainly in Switzerland and in the German-speaking world, concerning the organization of consumers and the planning of co-operative housing. Starting in the late 18th century, theorizations were developed to "crush the vice of capitalism", as Swiss economist Johann Friedrich Schär expressed it, when writing about Freidorf in 1921. In the fourth part, the monographic approach of the second part is mixed with the history of ideas in the third part, by means of a reconstruction of the genesis of Siedlung Freidorf. Archival analysis of the more than hundred board meetings of the cooperative - the first was held on May 20th, 1919 - structures a chronology of the problems that arose and the decisions that were taken, both organizational and architectural, during the construction of Freidorf, up until the official inauguration on June 1st, 1924.

The fifth and final part is an examination of the immediate reception, interpretation, and representation of the project, by means of its presence at the Exposition Internationale de la Cooperation et des Oeuvres Sociales by the International Cooperative Alliance, in the summer of 1924 in Ghent. Here, Meyer installed the Theater Co-op: an auditorium, a series of paintings, a glazed showcase with consumer goods, as well as a puppet theater performing the dream of cooperation and its realization in Freidorf. Architecture became propaganda, but the event was also the occasion for Meyer to look back - critically - on what he had achieved.

In the epilogue of the thesis, a set of interpretations is given of the well-known image "Co-op Interieur", produced by Hannes Meyer in 1926, shortly after the completion of Freidorf. In an appendix, six of his seminal articles relating to Freidorf and written between 1919 and 1924, are translated into English, as well as two texts by Johann Friedrich Schär, and Bernhard Jaeggi and Karl

Distributed systems, which enable the efficient processing of large-scale data and complex tasks across multiple interconnected processes, offer significant advantages over single-machine systems. These include enhanced scalability, improved fault tolerance, and the effective utilization of geographically distributed resources. A fundamental challenge in such systems lies in achieving coordination among unsynchronized processes that often operate over unreliable networks and must tolerate various types of failures. Achieving consensus -- agreement on a shared state or decision across processes -- is central to overcoming this challenge but remains a highly non-trivial problem. Consensus mechanisms underpin critical (distributed) services such as distributed key generation (DKG), secure multi-party computation (MPC), blockchain technologies, and state machine replication (SMR), emphasizing their indispensable role in modern computing.

Despite their significance, existing consensus protocols face notable limitations. Many struggle to scale efficiently due to poor performance or over-reliance on cryptographic tools such as threshold signatures, which demand costly setup and computationally intensive algebraic operations. Furthermore, reliance on these ``heavyweight'' cryptographic primitives compromises their security in a post-quantum world. As distributed systems continue to grow in scale and importance, there is a pressing need for consensus solutions that are not only efficient but also resilient to emerging security threats and future technological advancements.

This thesis addresses these challenges by revisiting two fundamental problems in distributed computing -- \emph{graded consensus} and \emph{Byzantine agreement}. We present the first complexity-optimal protocols for these problems that either leverage only lightweight'' cryptographic primitives, such as hash functions, or entirely eliminate the reliance on cryptography. Our solutions span all major network models -- \emph{asynchrony}, \emph{partial synchrony}, and \emph{synchrony}. By introducing algorithms that match or exceed the best known solutions, all while relying on little'' cryptography, this thesis advances the state of the art in consensus protocols and lays the foundation for more efficient and future-proof distributed systems capable of meeting the evolving demands of modern computing.

This thesis addresses key challenges in developing adaptive lower-limb exoskeletons to enhance balance, mobility, and user autonomy during daily living activities. It presents a comprehensive exploration of three critical aspects: push recovery during standing, partial assistance during gait and stair navigation, and real-time locomotion transition detection.

A novel push recovery framework integrates a bio-inspired stepping strategy and online optimization of step parameters, enabling natural and effective responses to external perturbations. Experimental validation demonstrates the framework's capability to enhance stability and support synergistic human-exoskeleton interaction.

To support daily living activities such as walking and stairs navigation, 3D path and flow controllers were developed, extending 2D implementations by incorporating hip abduction/adduction control to improve mediolateral stability. These controllers maintain natural movement variability while offering targeted balance assistance. Experimental results reveal that the path controller enhances trajectory alignment with inter-joint coordination patterns, while the flow controller provides intuitive and user-preferred support.

For locomotion transition detection, a machine learning-trained threshold-based method was introduced, achieving high real-time classification accuracy across two distinct exoskeletons, eWalk and Autonomyo. Personalization techniques, including Bayesian optimization, tailored the system to individual gait patterns, enhancing robustness and adaptability.

The findings emphasize the importance of incorporating dynamic balance mechanisms, intuitive control frameworks, and user-specific adaptability to address the complex demands of daily living activities while maintaining user autonomy and comfort. Future research should aim to integrate all these aspects, balance recovery, partial assistance during daily living activities, and transition detection, into cohesive frameworks. These frameworks should explore dynamic and real-world environments while prioritizing user-centric design approaches. Involving end-users, physiotherapists, and clinicians in the development process can further enhance that exoskeleton systems align with practical needs and preferences, advancing their potential to improve mobility and quality of life for individuals with walking impairments.

Multi-Object Tracking (MOT) is a fundamental computer vision task that involves detecting objects of interest in video frames and associating detections of the same object across time to form trajectories. For association, appearance cues, extracted through person re-identification (ReID) models, play a crucial role by capturing distinctive visual features of the tracked targets. However, despite its importance for tracking, ReID has primarily been studied as an image retrieval problem, with state-of-the-art methods overlooking tracking-specific challenges. This thesis focuses on advancing person re-identification methods, with a particular emphasis on making them more robust and better suited for tracking applications. A key challenge in ReID and MOT is handling occlusions, where targets become partially hidden by objects or other people, leading to degraded re-identification accuracy and potential identity switches in tracking. Additionally, tracking methods often fail to effectively combine ReID with motion cues and scene context, relying instead on naive association strategies. To address these challenges, this thesis makes three key contributions: BPBreID, a part-based method for robust occluded re-identification; KPR, a keypoint promptable ReID model designed to address multi-person occlusions scenarios; and CAMELTrack, an online tracking-by-detection method that replaces traditional heuristic for detection association with a context-aware learnable module. At the time of writing, KPR and CAMELTrack achieve state-of-the-art performance on widely-used benchmarks for occluded person re-identification and multi-object tracking. Complementary to this thesis work, additional research contributions were made in sports analytics, including the development of specialized re-identification models for athletes and the introduction of novel datasets for player tracking, re-identification, and jersey number recognition. The thesis concludes with a critical analysis of the current state of tracking and re-identification technologies, offering my opinionated view about the future of these two rapidly evolving fields.

The central aim of this thesis is to investigate single molecule phenomena in self-assembled plasmonic nanocavities, focusing on achieving single molecule sensitivity within these highly confined electromagnetic environments. While self-assembled nanocavities offer reproducible nanogaps and strong near-field enhancement, the collective contribution of multiple molecules often obscures single-molecule behavior. This work presents experimental results and electromagnetic simulations that demonstrate how single-molecule sensitivity can be achieved and applied.

The research explores various aspects of light-matter interactions at the nanoscale, including excitation enhancement, the Purcell effect and near-field coupling. Specific phenomena such as picocavity formation in nanoparticle-on-mirror structures and fluorescence reshaping in DNA origami-based nanocavities are studied. Additionally, the thesis examines how spectral overlap and dipole-dipole interactions between molecules can be enhanced in these nanocavities.