InfoscienceUnlocking Knowledge
Recent Scholarly Works
  • Some of the metrics are blocked by your 
    Publication

    On Drawing Architecture by Hand in the Age of Artificial Intelligence

    (Axel Menges, 2027-02) ;
    Schultz, Anne-Catrin

    This book makes the case for hand drawing as an essential, irreplaceable skill in the design process and one that artificial intelligence cannot substitute but must work alongside. It traces the act of drawing back to its roots as a mode of exploration and expression, arguing that architecture must be intuited »by hand« to preserve its humanity. The book’s plea for drawing by hand is grounded in architectural theory, the philosophy of learning, and neuroscience, examining the indispensable connection between brain and hand as a wellspring of human creativity. The freehand drawings created during the architectural design process frequently contain the essence of a project – a magic that unfolds through the ambiguity of the search for concept, function, and expression in a specific place. By letting lines remain tentative and exploratory, the space of a drawing opens room for expansion and alternatives, connecting the possible to the imaginary. Lines are marks. They are like memories that remain even after they’re erased. Hand-drawn lines record an evolving architecture through the successive layers and strata of design research. Fragile, imperfect, and ever-changing, like us, they mirror the creative process itself. The book is illustrated with a chronological sequence of architectural drawings spanning periods, geographies, and practices – from Renaissance studies to contemporary design experiments. Despite their diversity of context and authorship, these works share a common thread: drawing as a mode of thinking, not mere representation. Occasional imprecision, density, ambiguity, and layering expose the multiple reinterpretations and evolving insights inherent to the design process. Together, text and drawings reveal our collective culture and memory – a spatial intuition bound to identity, value, and narrative that emerges from a process-based, physical understanding of architecture grounded in actual making.

  • Some of the metrics are blocked by your 
    Dataset or other product

    Reducing Specimen Charging of Cryo Samples with Laser Illumination

    When a cryo sample is exposed to the electron beam, specimen charging can deflect the transmitted electrons and thus degrade the image quality. By heating the specimen with a laser, the charges that accumulate in the vitreous ice can be dissipated. Time-resolved experiments are used to determine the rate of charge dissipation. The results suggest that charge dissipation occurs as the sample is heated above 120 K, where the rearrangements of water molecules become possible, allowing protons trapped at defects to become mobile.

  • Some of the metrics are blocked by your 
    Dataset or other product

    pNEUMA

    (École Polytechnique Fédérale de Lausanne, 2023) ;

    The new era of sharing information and “big data” has raised our expectations to make mobility more predictable and controllable through a better utilization of data and existing resources. The realization of these opportunities requires going beyond the existing traditional ways of collecting traffic data that are based either on fixed-location sensors or GPS devices with low spatial coverage or penetration rates and significant measurement errors, especially in congested urban areas. Unmanned Aerial Systems (UAS) or simply “drones” have been proposed as a pioneering tool of the Intelligent Transportation Systems (ITS) infrastructure due to their unique characteristics, but various challenges have kept these efforts only at a small size. This dataset includes videos and trajectories from a series of drone experiments, nicknamed pNEUMA, to create the most complete urban dataset to study congestion. The aim of the experiment was to record traffic streams in a multi-modal congested environment over an urban setting using UAS that can allow the deep investigation of critical traffic phenomena. The pNEUMA experiments develop a prototype system that offers immense opportunities for researchers many of which are beyond the interests and expertise of traffic engineers. pNEUMA is part of an open science initiative aiming to create a unique observatory of traffic congestion, a scale an-order-of-magnitude higher than what was available till now, that researchers from different disciplines around the globe can use to develop and test their own models.

  • Some of the metrics are blocked by your 
    Publication

    Chirality Imprinting and Spin‐Texture Tunability in Conformally Coated 3D Magnetic Nanostructured Metamaterials

    (Wiley, 2026-11-18)
    Roberts, Alexander
    ;
    ;
    Askey, Joseph
    ;
    Veryard, Joseph
    ;
    Lanka, Satya

    Three‐dimensional (3D) magnetic nanostructures offer unprecedented opportunities for engineering emergent spin textures, but controlling their configuration remains a central challenge. Here, we show that conformally coated Ni nanotubes arranged in a woodpile geometry with lattice spacings ranging from 800 to 1200 nm, realized by two‐photon lithography and atomic layer deposition, exhibit a geometry‐tunable balance between chiral and axial states. Magnetic force microscopy on the top layer of the 3D woodpile reveals that few‐layer systems exhibit a chiral contrast, with an azimuthal magnetization wrapping around the tube circumference, whilst increasing the number of stacked layers drives a transition to an axial configuration, with the change in state populations depending strongly on lattice spacing. Micromagnetic simulations demonstrate that chirality is not intrinsic to isolated tubes but is imprinted by spin textures formed in the substrate sheet film, which couple into the 3D network. As the sheet‐film influence diminishes with increasing layer number, dipolar interactions dominate and stabilize the axial state. This two‐stage mechanism of chirality imprinting followed by increasingly dominant dipolar interactions provides clear control parameters for tailoring spin‐texture populations. Our results establish conformally coated woodpiles as a geometry‐programmable 3D magnetic metamaterial with applications in data storage, magnonics, and neuromorphic computing.

  • Some of the metrics are blocked by your 
    Publication

    Synthetic Data in MR Spectroscopy: Current Practices, Applications, and Considerations

    (Wiley, 2026-11-17)
    LaMaster, John T.
    ;
    Gudmundson, Aaron T.
    ;
    Abaei, Alireza
    ;
    Alcicek, Seyma
    ;
    Alvarado, Arturo

    Synthetic data generation has become an increasingly important component of methodological development in magnetic resonance spectroscopy (MRS), particularly in contexts where ground‐truth in vivo measurements are limited, noisy, or inaccessible. This review provides a structured overview of current practices in synthetic MRS data generation, with emphasis on the physical, biochemical, and computational assumptions that underpin commonly used models. A decomposition of synthetic MRS data is presented in terms of core building blocks, including basis sets, signal formation models, and metabolite parameter distributions such as concentration and relaxation properties. This framework is extended to advanced components required for realistic data synthesis, including baseline and nuisance signals, spatial encoding effects such as coil sensitivities and field inhomogeneities, k‐space sampling and gradient imperfections, as well as sample‐specific and temporal dynamics. Beyond single‐voxel modeling, synthetic data generation across multiple MRS modalities is discussed, including functional MRS, diffusion‐weighted MRS, and spectroscopic imaging, with emphasis on modality‐specific challenges such as physiological variability, motion, and confounding signal contributions. The role of synthetic data in clinical and preclinical research, acquisition optimization, software validation, and emerging artificial intelligence applications is further examined, with particular attention to domain shift and representational realism. Current practices in validation, reporting standards, and data accessibility are also reviewed, and key gaps limiting reproducibility and interoperability across studies are identified. Overall, this work provides a unified conceptual framework for synthetic MRS data generation, clarifies underlying assumptions across modeling choices, and outlines open challenges toward improving realism, standardization, and utility in both methodological and translational applications. Conducted by the MRS Synthetic Data Working Group under the Code & Data Sharing Committee of the MRS Study Group of the International Society for Magnetic Resonance in Medicine (ISMRM), this review encompasses the existing literature, supplemented by the collective experience and in‐house methodologies of its contributors, providing a community perspective that extends beyond the published record.

Recent EPFL Theses
  • Some of the metrics are blocked by your 
    Publication

    Multipanel Point-of-Care Electrochemical Biosensing Platform for Therapeutic Drug Monitoring in Precision Oncology

    Chemotherapy remains a cornerstone of cancer treatment, yet its clinical management is complicated by narrow therapeutic windows, pronounced inter-individual pharmacokinetic variability, and the routine co-administration of multiple drugs. These factors create a strong clinical need for therapeutic drug monitoring (TDM), which adjusts dosing to keep each patient within the therapeutic range. Current TDM relies on chromatographic and mass-spectrometric methods that are accurate but slow, costly, and confined to centralized laboratories, preventing the real-time, bedside feedback required for individualized dose optimization in oncology.

    This thesis develops an electrochemical biosensing platform for point-of-care TDM of six clinically relevant chemotherapeutic agents — cyclophosphamide, ifosfamide, etoposide, methotrexate, 5-fluorouracil, and busulfan — selected for their narrow therapeutic indices and high inter-individual variability. Because these drugs present a wide range of electrochemical behaviors, two complementary detection strategies were developed. Intrinsically electroactive drugs (etoposide, methotrexate, 5-fluorouracil) were detected directly at carbon electrodes; their simultaneous detection on a single sensor was demonstrated, and a previously uncharacterized mutual electrochemical cross-reactivity between etoposide and methotrexate was identified and mechanistically explained. Carboxylated multi-walled carbon nanotubes were further shown to selectively amplify the first oxidation peak of etoposide through an adsorption-mediated interfacial process.

    Non-electroactive drugs were addressed through enzyme-mediated transduction. Cytochrome P450 biosensors were built on nanostructured supports, comparing spinel oxide nanoparticles with platinum nanopetals; the latter enabled, to the best of our knowledge, the first demonstration of mediator-free direct electron transfer with CYP2B6, yielding a sensitivity exceeding previously reported CYP-based biosensors by more than one order of magnitude. For busulfan, which is neither electroactive nor a CYP substrate, the first inhibition-based electrochemical biosensor was realized through a glutathione S-transferase mechanism.

    These strategies were integrated into a dual-channel biosensor combining direct and enzymatic detection on a single substrate, enabling the detection of multiple drugs and their simultaneous measurement in clinically relevant combinations, without cross-interference between channels. The platform was validated in undiluted human serum, where matrix effects and electrode biofouling were mitigated using drug-specific strategies, including acetonitrile protein precipitation and a protective chitosan layer for the enzymatic electrodes. Finally, a dedicated, battery-powered electrochemical front-end supporting cyclic and differential pulse voltammetry, therapy-based operation, and wireless communication was designed and integrated onto a compact printed circuit board, reproducing benchtop-instrument performance in a portable form.

    Together, these contributions deliver a complete proof of concept for portable, multi-drug electrochemical TDM, a concrete step toward closed-loop, individualized chemotherapy management at the patient's bedside.

      8
  • Some of the metrics are blocked by your 
    Publication

    Dimension-Reduced Surrogates and Memory-Assisted Projections for Expensive Parametric and Evolution Problems

    Many scientific computing tasks repeatedly evaluate expensive models to quantify uncertainty, optimize a design, or approximate the solution of a dynamical system. Yet successive computations often share structure because they revisit either the same target map or nearby ones. This thesis investigates how information that has already been acquired can reduce the cost of the computations that follow. It is organized around two questions:

    • (1) When the target is fixed, how can expensive information be used to construct a reusable surrogate for later queries?
    • (2) When the target evolves along a trajectory, how can information from previous steps predict the approximation needed at the current one, so that only limited new expensive information must be computed?

    The first part of the thesis addresses fixed target maps by constructing global surrogates that can be evaluated repeatedly without calling the expensive model. We build on the active-subspace method, which identifies a few linear combinations of the input parameters along which the model output varies most. We prove that coordinates learned from finitely many gradient evaluations of the target map are, with high probability, nearly as accurate as the best reduced coordinates in the error measure relevant to the resulting surrogate. The number of required gradients is controlled by how concentrated the informative parameter directions are under the sampling distribution. Known smoothness of the parametric model then provides practical sampling rules.

    We then propose a multilevel version of active-subspace surrogate construction, assigning separate reduced coordinates and polynomial approximations to the corrections between successive model fidelities. This allocation uses many evaluations of inexpensive coarse models to capture the dominant behavior and reserves costly fine-model evaluations for the smaller details that remain. For an idealized construction, the analysis quantifies when this strategy reduces work compared with a single-level method. An adaptive implementation produces an evaluable surrogate, and numerical experiments demonstrate lower construction work at comparable accuracy.

    The second part addresses targets that evolve during optimization or time integration. For stochastic optimization, we propose a method that fits stored sample gradients by weighted least squares to predict the current objective gradient and uses a fresh sample-gradient difference to correct the prediction without introducing bias. The number of fresh gradients remains controlled throughout the iterations, thus past gradient evaluations replace most of the expensive new data that would otherwise be needed at each design.

    For the approximation of nonlinear dynamical systems we build on the Dirac-Frenkel variational principle, which evolves a parametrized state by fitting the system's vector field in the tangent space induced by the parametrization. We propose an inertial extension that retains the previous parameter velocity and solves a regularized tangent least-squares problem only for its correction. We establish well-posedness and a posteriori state-error estimates in continuous and discrete time. When consecutive velocities are sufficiently coherent, the method can use stronger regularization and smaller randomized sketches while retaining accuracy, thereby reducing the cost of the correction computed at each time step. This behavior is supported by numerical experiments.

      9
  • Some of the metrics are blocked by your 
    Publication

    Ultra-Miniaturized and Ultra-Low-Power CMOS Interfaces for In-Body Biosensing

    Implantable biomedical systems are transitioning from large, battery-powered single-function devices to networks of microscale, wirelessly powered sensor nodes that monitor the brain's chemical and electrical state from within. Concepts such as Neural Dust and Body Dust envision implant nodes the size of blood cells, harvesting power wirelessly within a nanowatt energy budget. Realizing this vision requires electrochemical interfaces capable of detecting neurotransmitters such as dopamine and glutamate in vivo, since these molecules govern neurological processes invisible to electrical recording alone. Conventional CMOS potentiostats, however, rely on analog transimpedance amplifiers and explicit analog-to-digital converters that typically consume microwatts and occupy tenths of a square millimeter, three or more orders of magnitude beyond what sub-mm implants allow. This thesis addresses that gap.

    The work is organized around a fully digital potentiostat architecture in which sensing, digitization, and closed-loop electrode potential control are unified into a single compact block based on a digital operational transconductance amplifier. This design, with co-designed supporting circuits and CMOS-compatible microfabricated probe electrodes, is validated from sensor modeling and individual circuit blocks to a complete system-level neurochemical sensing platform, and is further extended to neural signal acquisition through a miniaturized intravascular recording front-end.

    The thesis reports six principal contributions. First, a more reliable equivalent circuit model for electrochemical sensors incorporating a Butlerâ Volmer voltage-dependent current source and constant-phase elements at both electrode interfaces, addressing CMOS simulation errors of up to 60% in current and 110% in saturation-risk prediction. Second, the fully digital potentiostat, fabricated in TSMC 180 nm CMOS at 0.4 V, consuming only 1.2 nW in 260 µm2 (excluding reference and oscillator), the lowest power and smallest area ever reported to date. Third, a capacitor-less Dynamic Leakage Suppression ring oscillator achieving 69 µm and 7.9 nW at 21.89 kHz. Fourth, two complementary low-dropout regulators delivering 400 mV at quiescent currents of 7.3 and 21.5 nA in 1563.7 and 338.8 µm2, respectively. Fifth, CMOS-compatible foldable platinum microprobes integrated with the digital potentiostat to form a complete platform, achieving dopamine detection at 1383 pulses/µM sensitivity and 151 nM limit of detection, and enzymatic glutamate detection at 269 pulses/µM and 1.2 µM limit of detection. Sixth, a 0.0025 mm2 intravascular neural front-end combining a differential capacitive-feedback low-noise amplifier with a VCO-based frequency encoder, achieving 5.55 µVrms input-referred noise over the spike band at 11 µW for catheter-deployable recording with body-coupled communication.

    Together, these contributions demonstrate that neurochemical sensing can be performed by a fully integrated CMOS system within the nanowatt power and sub-mm2 area envelope of distributed implantable devices. The results establish a validated design methodology spanning sensor co-design, digital circuit architecture, power management, clock generation, microfabrication, and system integration, providing a reusable foundation for next-generation implantable biosensors and bringing the Body Dust vision closer to clinical realization.

      12
  • Some of the metrics are blocked by your 
    Publication

    Generative Artificial Intelligence for Accelerating Materials Discovery

    The rapid convergence of artificial intelligence and domain knowledge is fundamentally transforming computational chemistry and materials science. However, bridging the gap between passive structural evaluation, goal-directed generation, and fully autonomous discovery loops remains a critical challenge. This dissertation establishes a unified framework that seamlessly transitions from accurate forward property prediction to inverse target-driven materials design, culminating in self-evolving agentic systems for autonomous research.

    First, to overcome the topological limitations of conventional graph representations in forward property prediction, two domain-tailored neural network architectures are developed. Chapter 2 introduces Molecular Hypergraph Neural Networks (MHNN), which model higher-order delocalized connections in conjugated organic molecules, outperforming standard 2D and 3D graph neural networks in optoelectronic property prediction with high data efficiency. Extending forward modeling to complex solid-gas catalytic interfaces, Chapter 3 presents AdsMT, a multi-modal transformer integrating surface graph topology and adsorbate representations through cross-attention mechanisms. AdsMT bypasses explicit site-binding enumeration to deliver fast, interpretable, and uncertainty-calibrated predictions of global minimum adsorption energies across diverse catalyst systems.

    Building upon robust forward evaluators, Chapter 4 shifts the paradigm from forward property mapping to proactive inverse design. We present MatInvent, a reinforcement learning (RL) framework that fine-tunes pre-trained 3D crystal diffusion models for goal-directed materials synthesis. By framing reverse denoising as a Markov decision process with reward-weighted Kullback-Leibler regularization, experience replay, and diversity filtering, MatInvent achieves sample-efficient de novo generation of stable crystal structures tailored to both single-objective targets and complex multi-objective trade-offs (e.g., low-supply-risk magnets and high-k gate dielectrics).

    Finally, to unite isolated predictive and generative models into an end-to-end research ecosystem, Chapter 5 develops CASCADE, a self-evolving multi-agent framework that transitions large language models from static tool usage to cumulative skill acquisition. Driven by meta-skills of continuous learning and self-reflection, CASCADE autonomously conducts computational reasoning, hypothesis testing, and laboratory automation tasks. On a curated benchmark of 116 materials and chemistry tasks, CASCADE achieves state-of-the-art task execution success and demonstrates transferable skill consolidation across heterogeneous tasks and model backbones.

    In summary, this dissertation provides a logical progression, from capturing fundamental structure-property relationships to inverse-designing target materials, and ultimately orchestrating closed-loop automated research, offering a scalable foundation for modern AI-driven scientific discovery.

      2
  • Some of the metrics are blocked by your 
    Publication

    Bayesian Plasma Tomography in Tokamak Fusion Devices: A Unifying Framework, Real-Time Inference, and Learned Priors

    Nuclear fusion promises a low-carbon and potentially near-limitless energy source. Currently, the leading approach to fusion power is the magnetic confinement of high-temperature ionized gases called plasmas, in devices such as tokamaks and stellarators. Present-day experiments aim to understand and solve the physics and engineering problems that stand in the way, relying on diagnostics that probe the plasma's properties and performance. Among these are systems measuring the radiation emitted by the plasma in different energy ranges, such as soft X-ray detectors or broadband bolometers. To a first approximation, these detectors measure the plasma radiation integrated along their lines-of-sight. Tomography then reconstructs the spatial emissivity distribution from these data. Technical and design constraints, however, drastically limit the number of measurements. Estimating the emissivity profiles is thus a severely underdetermined problem: for a given set of measurements, infinitely many profiles are perfectly consistent with the data.

    Bayesian statistics provides a systematic and principled way to account for the resulting uncertainty. Statistical modeling of the observed data leads to a likelihood term, while the assumed properties of the profile are encoded in a prior term; together, they yield the posterior, which models all the available information about the profile. The prior compensates for the missing information, at the price of biasing the solution.

    In this thesis, we develop a unifying Bayesian framework for plasma tomography, framing the problem in the language of modern computational imaging. Using robust optimization and sampling algorithms, we show how to extract uncertainty-aware estimates of quantities of interest from the posterior; the most popular tomography techniques are recovered as deterministic point estimates within this framework. We then compute, for the TCV tokamak at EPFL, physically accurate volume-of-sight models of the diagnostic views; these improve the overall reconstruction while barely affecting the estimated power radiated by large plasma regions, already well captured by simpler line-of-sight models. We propose a technique providing accurate real-time estimates of the radiated power as an analytically derived linear combination of bolometer measurements. Its coefficients are pre-computed before a given discharge, allowing scenario-dependent adaptation and tracking radiation from arbitrary regions of interest. We report on the first TCV experiments to rely on real-time bolometry, where plasma exhaust was actively controlled with our technique. Finally, we investigate data-driven priors based on diffusion models, learned from physically realistic profiles. The resulting model, despite not relying on the magnetic equilibrium, can outperform hand-designed equilibrium-informed priors, and its estimated uncertainty better localizes the reconstruction error.

    All the developed approaches are facets of the same general Bayesian framework. Given the extremely underdetermined nature of plasma tomography, none of them can be expected to be flawless: robustness, accuracy, speed, interpretability and model expressiveness cannot be maximized simultaneously. We explore these fundamental trade-offs and argue that a sound treatment of this complex problem should combine different probabilistic methods, benefiting from the best features of each. All the developed computational routines are made openly available.