arXiv subjects

Tonio Ball

Publications and source records attributed to Tonio Ball.

At least 19 recordsLinked to original sources

arXiv2026

Neuro-Inspired Inverse Learning for Planning and Control

We present a neuro-inspired framework for embodied planning and control. Building on three principles that enable fast and highly effective goal-directed behavior in the mammalian brain - paired forward/inverse internal models, open-loop multi-step motor commands, and sequential, hierarchical organization of action - our Inverter framework uses learned components, trained end-to-end through Inverse Learning (IL) and supplemented where natural by analytic or algorithmic modules; we formalize IL and delineate it from supervised, reinforcement, and imitation learning. IL bridges Reinforcement Learning (RL)-style amortization, which runs in a single forward pass but emits only one action at a time, and Optimal Control (OC)-style sequence planning over whole trajectories, but with iterative test-time computation. Single Inverters or hierarchical n=2 Inverter stacks match or improve on offline-RL and diffusion-planner baselines on all 3 maze2d and 6 antmaze D4RL variants by an average of +24.2% (range -1.9% to +78.2%), at one-to-two orders of magnitude less inference compute time. Distinctively, optimizing through the Forward Model (FoM) over the entire T-step action sequence - rather than per step - lets Inverters produce smooth, goal-coherent, trajectory-wide structure and reach control policies closer to the analytic optimum than the policy underlying the training data itself. We also identify a failure mode of IL: FoM hacking under narrow training-data coverage, which we mitigate by using random training data with broader coverage. As an application example, a Pulse Inverter synthesizes arbitrary single-qubit quantum gates with fidelity matching the standard iterative numerical baseline (GRAPE), at more than 1000x lower per-gate compute time. In summary, we conclude that IL enables a versatile class of world-interfaces, especially for latency- and resource-critical embodied AI.

cs.AI↗

arXiv2026

Human Cardiac Measurements with Diamond Magnetometers

We demonstrate direct, non-invasive and non-contact detection of human cardiac magnetic signals using quantum sensors based on nitrogen-vacancy (NV) centers in diamond. Three configurations were employed recording magnetocardiography (MCG) signals in various shielded and unshielded environments. The signals were averaged over a few hundreds up to several thousands of heart beats to detect the MCG traces. The compact room-temperature NV sensors exhibit sensitivities of 6-26 pT/Hz^(1/2) with active sensing volumes below 0.5 mm^3, defining the performance level of the demonstrated MCG measurements. While the present signals are obtained by averaging, this performance already indicates a clear path toward single-shot MCG sensing. To move beyond shielded environments toward practical clinical use, strong noise suppression is required. To this end, we implement NV-based gradiometry and achieve efficient common-mode noise rejection, enabled by the intrinsically small sensing volume of NV sensors. Together, these multi-platform results obtained across diverse magnetic environments provide a solid foundation for translating quantum sensors into human medical diagnostics such as MCG and magnetoencephalography (MEG).

physics.med-ph↗

arXiv2025

EEG-CLIP : Learning EEG representations from natural language descriptions

Deep networks for electroencephalogram (EEG) decoding are often only trained to solve one specific task, such as pathology or age decoding. A more general task-agnostic approach is to train deep networks to match a (clinical) EEG recording to its corresponding textual medical report and vice versa. This approach was pioneered in the computer vision domain matching images and their text captions and subsequently allowed to do successful zero-shot decoding using textual class prompts. In this work, we follow this approach and develop a contrastive learning framework, EEG-CLIP, that aligns the EEG time series and the descriptions of the corresponding clinical text in a shared embedding space. We investigated its potential for versatile EEG decoding, evaluating performance in a range of few-shot and zero-shot settings. Overall, we show that EEG-CLIP manages to non-trivially align text and EEG representations. Our work presents a promising approach to learn general EEG representations, which could enable easier analyses of diverse decoding questions through zero-shot decoding or training task-specific models from fewer training examples. The code for reproducing our results is available at https://github.com/tidiane-camaret/EEGClip

cs.CL↗

arXiv2025

Unlocking In-Context Learning for Natural Datasets Beyond Language Modelling

Large Language Models (LLMs) exhibit In-Context Learning (ICL), which enables the model to perform new tasks conditioning only on the examples provided in the context without updating the model's weights. While ICL offers fast adaptation across natural language tasks and domains, its emergence is less straightforward for modalities beyond text. In this work, we systematically uncover properties present in LLMs that support the emergence of ICL for autoregressive models and various modalities by promoting the learning of the needed mechanisms for ICL. We identify exact token repetitions in the training data sequences as an important factor for ICL. Such repetitions further improve stability and reduce transiency in ICL performance. Moreover, we emphasise the significance of training task difficulty for the emergence of ICL. Finally, by applying our novel insights on ICL emergence, we unlock ICL capabilities for various visual datasets and a more challenging EEG classification task.

cs.CL↗

arXiv2024

Human-AI Teaming Using Large Language Models: Boosting Brain-Computer Interfacing (BCI) and Brain Research

Recently, there is an increasing interest in using artificial intelligence (AI) to automate aspects of the research process, or even autonomously conduct the full research cycle from idea generation, over data analysis, to composing and evaluation of scientific manuscripts. Examples of working AI scientist systems have been demonstrated for computer science tasks and running molecular biology labs. While some approaches aim for full autonomy of the scientific AI, others rather aim for leveraging human-AI teaming. Here, we address how to adapt such approaches for boosting Brain-Computer Interface (BCI) development, as well as brain research resp. neuroscience at large. We argue that at this time, a strong emphasis on human-AI teaming, in contrast to fully autonomous AI BCI researcher will be the most promising way forward. We introduce the collaborative workspaces concept for human-AI teaming based on a set of Janusian design principles, looking both ways, to the human as well as to the AI side. Based on these principles, we present ChatBCI, a Python-based toolbox for enabling human-AI collaboration based on interaction with Large Language Models (LLMs), designed for BCI research and development projects. We show how ChatBCI was successfully used in a concrete BCI project on advancing motor imagery decoding from EEG signals. Our approach can be straightforwardly extended to broad neurotechnological and neuroscientific topics, and may by design facilitate human expert knowledge transfer to scientific AI systems in general.

cs.HC↗

arXiv2023

Brain Age Revisited: Investigating the State vs. Trait Hypotheses of EEG-derived Brain-Age Dynamics with Deep Learning

The brain's biological age has been considered as a promising candidate for a neurologically significant biomarker. However, recent results based on longitudinal magnetic resonance imaging data have raised questions on its interpretation. A central question is whether an increased biological age of the brain is indicative of brain pathology and if changes in brain age correlate with diagnosed pathology (state hypothesis). Alternatively, could the discrepancy in brain age be a stable characteristic unique to each individual (trait hypothesis)? To address this question, we present a comprehensive study on brain aging based on clinical EEG, which is complementary to previous MRI-based investigations. We apply a state-of-the-art Temporal Convolutional Network (TCN) to the task of age regression. We train on recordings of the Temple University Hospital EEG Corpus (TUEG) explicitly labeled as non-pathological and evaluate on recordings of subjects with non-pathological as well as pathological recordings, both with examinations at a single point in time and repeated examinations over time. Therefore, we created four novel subsets of TUEG that include subjects with multiple recordings: I) all labeled non-pathological; II) all labeled pathological; III) at least one recording labeled non-pathological followed by at least one recording labeled pathological; IV) similar to III) but with opposing transition (first pathological then non-pathological). The results show that our TCN reaches state-of-the-art performance in age decoding with a mean absolute error of 6.6 years. Our extensive analyses demonstrate that the model significantly underestimates the age of non-pathological and pathological subjects (-1 and -5 years, paired t-test, p <= 0.18 and p <= 0.0066). Furthermore, the brain age gap biomarker is not indicative of pathological EEG.

q-bio.NC↗

arXiv2022

Deep Riemannian Networks for End-to-End EEG Decoding

State-of-the-art performance in electroencephalography (EEG) decoding tasks is currently often achieved with either Deep-Learning (DL) or Riemannian-Geometry-based decoders (RBDs). Recently, there is growing interest in Deep Riemannian Networks (DRNs) possibly combining the advantages of both previous classes of methods. However, there are still a range of topics where additional insight is needed to pave the way for a more widespread application of DRNs in EEG. These include architecture design questions such as network size and end-to-end ability. How these factors affect model performance has not been explored. Additionally, it is not clear how the data within these networks is transformed, and whether this would correlate with traditional EEG decoding. Our study aims to lay the groundwork in the area of these topics through the analysis of DRNs for EEG with a wide range of hyperparameters. Networks were tested on five public EEG datasets and compared with state-of-the-art ConvNets. Here we propose EE(G)-SPDNet, and we show that this wide, end-to-end DRN can outperform the ConvNets, and in doing so use physiologically plausible frequency regions. We also show that the end-to-end approach learns more complex filters than traditional band-pass filters targeting the classical alpha, beta, and gamma frequency bands of the EEG, and that performance can benefit from channel specific filtering approaches. Additionally, architectural analysis revealed areas for further improvement due to the possible under utilisation of Riemannian specific information throughout the network. Our study thus shows how to design and train DRNs to infer task-related information from the raw EEG without the need of handcrafted filterbanks and highlights the potential of end-to-end DRNs such as EE(G)-SPDNet for high-performance EEG decoding.

cs.LG↗

arXiv2022

A Framework for Preserving Privacy and Cybersecurity in Brain-Computer Interfacing Applications

Brain-Computer Interfaces (BCIs) comprise a rapidly evolving field of technology with the potential of far-reaching impact in domains ranging from medical over industrial to artistic, gaming, and military. Today, these emerging BCI applications are typically still at early technology readiness levels, but because BCIs create novel, technical communication channels for the human brain, they have raised privacy and security concerns. To mitigate such risks, a large body of countermeasures has been proposed in the literature, but a general framework is lacking which would describe how privacy and security of BCI applications can be protected by design, i.e., already as an integral part of the early BCI design process, in a systematic manner, and allowing suitable depth of analysis for different contexts such as commercial BCI product development vs. academic research and lab prototypes. Here we propose the adoption of recent systems-engineering methodologies for privacy threat modeling, risk assessment, and privacy engineering to the BCI field. These methodologies address privacy and security concerns in a more systematic and holistic way than previous approaches, and provide reusable patterns on how to move from principles to actions. We apply these methodologies to BCI and data flows and derive a generic, extensible, and actionable framework for brain-privacy-preserving cybersecurity in BCI applications. This framework is designed for flexible application to the wide range of current and future BCI applications. We also propose a range of novel privacy-by-design features for BCIs, with an emphasis on features promoting BCI transparency as a prerequisite for informational self-determination of BCI users, as well as design features for ensuring BCI user autonomy. We anticipate that our framework will contribute to the development of privacy-respecting, trustworthy BCI technologies.

cs.CR↗

arXiv2022

The spatial scale dimension of speech processing in the human brain

In the past three decades, neuroimaging has provided important insights into structure-function relationships in the human brain. Recently, however, the methods for analyzing functional magnetic resonance imaging (fMRI) data have come under scrutiny, with studies questioning cross-software comparability, the validity of statistical inference and interpretation, and the influence of the spatial filter size on neuroimaging analyses. As most fMRI studies only use a single filter for analysis, much information on the size and shape of the BOLD signal in Gaussian scale space remains hidden and constrains the interpretation of fMRI studies. To investigate the influence of the spatial observation scale on fMRI analysis, we use a spatial multiscale analysis with a range of Gaussian filters from 1-20 mm (full width at half maximum) to analyze fMRI data from a speech repetition paradigm in 25 healthy subjects. We show that analyzing the fMRI data over a range of Gaussian filter kernels reveals substantial variability in the neuroanatomical localization and the average signal strength and size of suprathreshold clusters depending on the filter size. We also demonstrate how small spatial filters bias the results towards subcortical and cerebellar clusters. Furthermore, we describe substantially different scale-dependent cluster size dynamics between cortical and cerebellar clusters. We discuss how spatial multiscale analysis may substantially improve the interpretation of fMRI data. We propose to further develop a spatial multiscale analysis to fully explore the deep structure of the BOLD signal in Gaussian scale space.

q-bio.NC↗

arXiv2022

When less is more: Simplifying inputs aids neural network understanding

How do neural network image classifiers respond to simpler and simpler inputs? And what do such responses reveal about the learning process? To answer these questions, we need a clear measure of input simplicity (or inversely, complexity), an optimization objective that correlates with simplification, and a framework to incorporate such objective into training and inference. Lastly we need a variety of testbeds to experiment and evaluate the impact of such simplification on learning. In this work, we measure simplicity with the encoding bit size given by a pretrained generative model, and minimize the bit size to simplify inputs in training and inference. We investigate the effect of such simplification in several scenarios: conventional training, dataset condensation and post-hoc explanations. In all settings, inputs are simplified along with the original classification task, and we investigate the trade-off between input simplicity and task performance. For images with injected distractors, such simplification naturally removes superfluous information. For dataset condensation, we find that inputs can be simplified with almost no accuracy degradation. When used in post-hoc explanation, our learning-based simplification approach offers a valuable new tool to explore the basis of network decisions.

cs.LG↗

arXiv2021

Towards a Governance Framework for Brain Data

The increasing availability of brain data within and outside the biomedical field, combined with the application of artificial intelligence (AI) to brain data analysis, poses a challenge for ethics and governance. We identify distinctive ethical implications of brain data acquisition and processing, and outline a multi-level governance framework. This framework is aimed at maximizing the benefits of facilitated brain data collection and further processing for science and medicine whilst minimizing risks and preventing harmful use. The framework consists of four primary areas of regulatory intervention: binding regulation, ethics and soft law, responsible innovation, and human rights.

q-bio.NC↗

arXiv2020

Understanding Anomaly Detection with Deep Invertible Networks through Hierarchies of Distributions and Features

Deep generative networks trained via maximum likelihood on a natural image dataset like CIFAR10 often assign high likelihoods to images from datasets with different objects (e.g., SVHN). We refine previous investigations of this failure at anomaly detection for invertible generative networks and provide a clear explanation of it as a combination of model bias and domain prior: Convolutional networks learn similar low-level feature distributions when trained on any natural image dataset and these low-level features dominate the likelihood. Hence, when the discriminative features between inliers and outliers are on a high-level, e.g., object shapes, anomaly detection becomes particularly challenging. To remove the negative impact of model bias and domain prior on detecting high-level differences, we propose two methods, first, using the log likelihood ratios of two identical models, one trained on the in-distribution data (e.g., CIFAR10) and the other one on a more general distribution of images (e.g., 80 Million Tiny Images). We also derive a novel outlier loss for the in-distribution network on samples from the more general distribution to further improve the performance. Secondly, using a multi-scale model like Glow, we show that low-level features are mainly captured at early scales. Therefore, using only the likelihood contribution of the final scale performs remarkably well for detecting high-level feature differences of the out-of-distribution and the in-distribution. This method is especially useful if one does not have access to a suitable general distribution. Overall, our methods achieve strong anomaly detection performance in the unsupervised setting, and only slightly underperform state-of-the-art classifier-based methods in the supervised setting. Code can be found at https://github.com/boschresearch/hierarchical_anomaly_detection.

cs.LG↗

arXiv2020

Machine-Learning-Based Diagnostics of EEG Pathology

Machine learning (ML) methods have the potential to automate clinical EEG analysis. They can be categorized into feature-based (with handcrafted features), and end-to-end approaches (with learned features). Previous studies on EEG pathology decoding have typically analyzed a limited number of features, decoders, or both. For a I) more elaborate feature-based EEG analysis, and II) in-depth comparisons of both approaches, here we first develop a comprehensive feature-based framework, and then compare this framework to state-of-the-art end-to-end methods. To this aim, we apply the proposed feature-based framework and deep neural networks including an EEG-optimized temporal convolutional network (TCN) to the task of pathological versus non-pathological EEG classification. For a robust comparison, we chose the Temple University Hospital (TUH) Abnormal EEG Corpus (v2.0.0), which contains approximately 3000 EEG recordings. The results demonstrate that the proposed feature-based decoding framework can achieve accuracies on the same level as state-of-the-art deep neural networks. We find accuracies across both approaches in an astonishingly narrow range from 81--86\%. Moreover, visualizations and analyses indicated that both approaches used similar aspects of the data, e.g., delta and theta band power at temporal electrode locations. We argue that the accuracies of current binary EEG pathology decoders could saturate near 90\% due to the imperfect inter-rater agreement of the clinical labels, and that such decoders are already clinically useful, such as in areas where clinical EEG experts are rare. We make the proposed feature-based framework available open source and thus offer a new tool for EEG machine learning research.

eess.IV↗

arXiv2019

A Research Framework for Virtual Reality Neurosurgery Based on Open-Source Tools

Fully immersive virtual reality (VR) has the potential to improve neurosurgical planning. For example, it may offer 3D visualizations of relevant anatomical structures with complex shapes, such as blood vessels and tumors. However, there is a lack of research tools specifically tailored for this area. We present a research framework for VR neurosurgery based on open-source tools and preliminary evaluation results. We showcase the potential of such a framework using clinical data of two patients and research data of one subject. As a first step toward practical evaluations, two certified senior neurosurgeons positively assessed the usefulness of the VR visualizations using head-mounted displays. The methods and findings described in our study thus provide a foundation for research and development aiming at versatile and user-friendly VR tools for improving neurosurgical planning and training.

cs.HC↗

arXiv2019

Deep Invertible Networks for EEG-based brain-signal decoding

In this manuscript, we investigate deep invertible networks for EEG-based brain signal decoding and find them to generate realistic EEG signals as well as classify novel signals above chance. Further ideas for their regularization towards better decoding accuracies are discussed.

cs.LG↗

arXiv2019

Cortical Mirror-System Activation During Real-Life Game Playing: An Intracranial Electroencephalography (EEG) Study

Analogous to the mirror neuron system repeatedly described in monkeys as a possible substrate for imitation learning and/or action understanding, a neuronal execution/observation matching system (OEMS) is assumed in humans, but little is known to what extent this system is activated in non-experimental, real-life conditions. In the present case study, we investigated brain activity of this system during natural, non-experimental motor behavior as it occurred during playing of the board game "Malefiz". We compared spectral modulations of the high-gamma band related to ipsilateral reaching movement execution and observation of the same kind of movement using electrocorticography (ECoG) in one participant. Spatially coincident activity during both conditions execution and observation was recorded at electrode contacts over the premotor/primary motor cortex. The topography and amplitude of the high-gamma modulations related to both, movement observation and execution were clearly spatially correlated over several fronto-parietal brain areas. Thus, our findings indicate that a network of cortical areas contributes to the human OEMS, beyond primary/premotor cortex including Brocas area and the temporo-parieto-occipital junction area, in real-life conditions.

q-bio.NC↗

arXiv2018

Intracranial Error Detection via Deep Learning

Deep learning techniques have revolutionized the field of machine learning and were recently successfully applied to various classification problems in noninvasive electroencephalography (EEG). However, these methods were so far only rarely evaluated for use in intracranial EEG. We employed convolutional neural networks (CNNs) to classify and characterize the error-related brain response as measured in 24 intracranial EEG recordings. Decoding accuracies of CNNs were significantly higher than those of a regularized linear discriminant analysis. Using time-resolved deep decoding, it was possible to classify errors in various regions in the human brain, and further to decode errors over 200 ms before the actual erroneous button press, e.g., in the precentral gyrus. Moreover, deeper networks performed better than shallower networks in distinguishing correct from error trials in all-channel decoding. In single recordings, up to 100 % decoding accuracy was achieved. Visualization of the networks' learned features indicated that multivariate decoding on an ensemble of channels yields related, albeit non-redundant information compared to single-channel decoding. In summary, here we show the usefulness of deep learning for both intracranial error decoding and mapping of the spatio-temporal structure of the human error processing network.

cs.LG↗

arXiv2018

Deep Learning for micro-Electrocorticographic (μECoG) Data

Machine learning can extract information from neural recordings, e.g., surface EEG, ECoG and μECoG, and therefore plays an important role in many research and clinical applications. Deep learning with artificial neural networks has recently seen increasing attention as a new approach in brain signal decoding. Here, we apply a deep learning approach using convolutional neural networks to μECoG data obtained with a wireless, chronically implanted system in an ovine animal model. Regularized linear discriminant analysis (rLDA), a filter bank component spatial pattern (FBCSP) algorithm and convolutional neural networks (ConvNets) were applied to auditory evoked responses captured by μECoG. We show that compared with rLDA and FBCSP, significantly higher decoding accuracy can be obtained by ConvNets trained in an end-to-end manner, i.e., without any predefined signal features. Deep learning thus proves a promising technique for μECoG-based brain-machine interfacing applications.

eess.SP↗