arXiv subjects

Gideon Vos

Publications and source records attributed to Gideon Vos.

18 recordsLinked to original sources

arXiv2025

Coarse-graining in time, emergent temperature in 2d CFT at large central charge

Using the monodromy method, a compact expression is obtained for the identity block contribution to the expectation value of two low-energy probe operators on a broad class of time-dependent heavy pure states in large-$c$ 2d CFTs. It will be shown that the size of the compact spatial dimension sets a coarse-graining time-scale above which the thermal two-point function naturally emerges. A phenomena will be highlighted where the information of the pure state appears to get projected down to a single conformal representation. We will see that this projection is corroborated by known conformal bootstrap results.

hep-th↗

arXiv2025

Synaptic Pruning: A Biological Inspiration for Deep Learning Regularization

Synaptic pruning in biological brains removes weak connections to improve efficiency. In contrast, dropout regularization in artificial neural networks randomly deactivates neurons without considering activity-dependent pruning. We propose a magnitude-based synaptic pruning method that better reflects biology by progressively removing low-importance connections during training. Integrated directly into the training loop as a dropout replacement, our approach computes weight importance from absolute magnitudes across layers and applies a cubic schedule to gradually increase global sparsity. At fixed intervals, pruning masks permanently remove low-importance weights while maintaining gradient flow for active ones, eliminating the need for separate pruning and fine-tuning phases. Experiments on multiple time series forecasting models including RNN, LSTM, and Patch Time Series Transformer across four datasets show consistent gains. Our method ranked best overall, with statistically significant improvements confirmed by Friedman tests (p < 0.01). In financial forecasting, it reduced Mean Absolute Error by up to 20% over models with no or standard dropout, and up to 52% in select transformer models. This dynamic pruning mechanism advances regularization by coupling weight elimination with progressive sparsification, offering easy integration into diverse architectures. Its strong performance, especially in financial time series forecasting, highlights its potential as a practical alternative to conventional dropout techniques.

cs.LG↗

arXiv2025

Decoding Neural Emotion Patterns through Large Language Model Embeddings

Understanding how emotional expression in language relates to brain function is a challenge in computational neuroscience and affective computing. Traditional neuroimaging is costly and lab-bound, but abundant digital text offers new avenues for emotion-brain mapping. Prior work has largely examined neuroimaging-based emotion localization or computational text analysis separately, with little integration. We propose a computational framework that maps textual emotional content to anatomically defined brain regions without requiring neuroimaging. Using OpenAI's text-embedding-ada-002, we generate high-dimensional semantic representations, apply dimensionality reduction and clustering to identify emotional groups, and map them to 18 brain regions linked to emotional processing. Three experiments were conducted: i) analyzing conversational data from healthy vs. depressed subjects (DIAC-WOZ dataset) to compare mapping patterns, ii) applying the method to the GoEmotions dataset and iii) comparing human-written text with large language model (LLM) responses to assess differences in inferred brain activation. Emotional intensity was scored via lexical analysis. Results showed neuroanatomically plausible mappings with high spatial specificity. Depressed subjects exhibited greater limbic engagement tied to negative affect. Discrete emotions were successfully differentiated. LLM-generated text matched humans in basic emotion distribution but lacked nuanced activation in empathy and self-referential regions (medial prefrontal and posterior cingulate cortex). This cost-effective, scalable approach enables large-scale analysis of naturalistic language, distinguishes between clinical populations, and offers a brain-based benchmark for evaluating AI emotional expression.

cs.CL↗

arXiv2025

A Statistical Approach for Synthetic EEG Data Generation

Electroencephalogram (EEG) data is crucial for diagnosing mental health conditions but is costly and time-consuming to collect at scale. Synthetic data generation offers a promising solution to augment datasets for machine learning applications. However, generating high-quality synthetic EEG that preserves emotional and mental health signals remains challenging. This study proposes a method combining correlation analysis and random sampling to generate realistic synthetic EEG data. We first analyze interdependencies between EEG frequency bands using correlation analysis. Guided by this structure, we generate synthetic samples via random sampling. Samples with high correlation to real data are retained and evaluated through distribution analysis and classification tasks. A Random Forest model trained to distinguish synthetic from real EEG performs at chance level, indicating high fidelity. The generated synthetic data closely match the statistical and structural properties of the original EEG, with similar correlation coefficients and no significant differences in PERMANOVA tests. This method provides a scalable, privacy-preserving approach for augmenting EEG datasets, enabling more efficient model training in mental health research.

eess.SP↗

arXiv2024

Signatures of Bulk Black Hole Merger from Semi-classical 2d CFT

We study the real-time bulk AdS$_3$ two-into-two scattering amplitude of conical defects. This is done through means of a previously developed method of systematic approximation of the monodromy problem for the Virasoro blocks of the four-point HHHH correlator in 2d CFT in the large central charge regime. We find that dialing the external scaling dimension triggers a transition from a scattering phase to an intermediate black hole phase. This transition occurs before the individual heavy operators exceed the BTZ mass gap.

hep-th↗

arXiv2024

Stabilizing Machine Learning for Reproducible and Explainable Results: A Novel Validation Approach to Subject-Specific Insights

Machine Learning is transforming medical research by improving diagnostic accuracy and personalizing treatments. General ML models trained on large datasets identify broad patterns across populations, but their effectiveness is often limited by the diversity of human biology. This has led to interest in subject-specific models that use individual data for more precise predictions. However, these models are costly and challenging to develop. To address this, we propose a novel validation approach that uses a general ML model to ensure reproducible performance and robust feature importance analysis at both group and subject-specific levels. We tested a single Random Forest (RF) model on nine datasets varying in domain, sample size, and demographics. Different validation techniques were applied to evaluate accuracy and feature importance consistency. To introduce variability, we performed up to 400 trials per subject, randomly seeding the ML algorithm for each trial. This generated 400 feature sets per subject, from which we identified top subject-specific features. A group-specific feature importance set was then derived from all subject-specific results. We compared our approach to conventional validation methods in terms of performance and feature importance consistency. Our repeated trials approach, with random seed variation, consistently identified key features at the subject level and improved group-level feature importance analysis using a single general model. Subject-specific models address biological variability but are resource-intensive. Our novel validation technique provides consistent feature importance and improved accuracy within a general ML model, offering a practical and explainable alternative for clinical research.

cs.LG↗

arXiv2024

The Effect of Acute Stress on the Interpretability and Generalization of Schizophrenia Predictive Machine Learning Models

Introduction Schizophrenia is a severe mental disorder, and early diagnosis is key to improving outcomes. Its complexity makes predicting onset and progression challenging. EEG has emerged as a valuable tool for studying schizophrenia, with machine learning increasingly applied for diagnosis. This paper assesses the accuracy of ML models for predicting schizophrenia and examines the impact of stress during EEG recording on model performance. We integrate acute stress prediction into the analysis, showing that overlapping conditions like stress during recording can negatively affect model accuracy. Methods Four XGBoost models were built: one for stress prediction, two to classify schizophrenia (at rest and task), and a model to predict schizophrenia for both conditions. XAI techniques were applied to analyze results. Experiments tested the generalization of schizophrenia models using their datasets' healthy controls and independent health-screened controls. The stress model identified high-stress subjects, who were excluded from further analysis. A novel method was used to adjust EEG frequency band power to remove stress artifacts, improving predictive model performance. Results Our results show that acute stress vary across EEG sessions, affecting model performance and accuracy. Generalization improved once these varying stress levels were considered and compensated for during model training. Our findings highlight the importance of thorough health screening and management of the patient's condition during the process. Stress induced during or by the EEG recording can adversely affect model generalization. This may require further preprocessing of data by treating stress as an additional physiological artifact. Our proposed approach to compensate for stress artifacts in EEG data used for training models showed a significant improvement in predictive performance.

eess.SP↗

arXiv2024

Stress Monitoring Using Low-Cost Electroencephalogram Devices: A Systematic Literature Review

Introduction. Low-cost health monitoring devices are increasingly being used for mental health related studies including stress. While cortisol response magnitude remains the gold standard indicator for stress assessment, a growing number of studies have started to use low-cost EEG devices as primary recorders of biomarker data. Methods. This study reviews published works contributing and/or using EEG devices for detecting stress and their associated machine learning methods. The reviewed works are selected to answer three general research questions and are then synthesized into four categories of stress assessment using EEG, low-cost EEG devices, available datasets for EEG-based stress measurement, and machine learning techniques for EEG-based stress measurement. Results. A number of studies were identified where low-cost EEG devices were utilized to record brain function during phases of stress and relaxation. These studies generally reported a high predictive accuracy rate, verified using a number of different machine learning validation methods and statistical approaches. Of these studies, 60% can be considered low-powered studies based on the small number of test subjects used during experimentation. Conclusion. Low-cost consumer grade wearable devices including EEG and wrist-based monitors are increasingly being used in stress-related studies. Standardization of EEG signal processing and importance of sensor location still requires further study, and research in this area will continue to provide improvements as more studies become available.

cs.HC↗

arXiv2023

Behind-the-horizon excitations from a single 2d CFT

In this work, we consider the atypical non-equilibrium state found in [1708.06328] which holographically represents a behind-the-horizon excitation in a blackhole spacetime. The special feature of this state is that it looks like an equilibrium state when probed by a class of low-energy operators. First, we retrieve this property using the uniformization mapping in the limit of large central charge, in the process we are able to derive rather than presume approximate thermal physics. Furthermore, in the large-c and high-energy limit we realize these excitations as elements of the commutant algebra of a GNS-representation of the light operator algebra. Instead of analytically continuing a mixed heavy-light Euclidean correlator to a Lorentzian correlator, we identify the Euclidean correlator as a GNS-linear form and interpret the Lorentzian correlator as a vacuum expectation value of representatives of the light operator algebra on the GNS-vacuum.

hep-th↗

arXiv2022

Holography for bulk states in 3D quantum gravity

In this work we discuss the holographic description of states in the Hilbert space of (2+1)-dimensional quantum gravity, living on a time slice in the bulk. We focus on pure gravity coupled to pointlike sources for heavy spinning particles. We develop a formulation where the equations for the backreacted metric reduce to two decoupled Liouville equations with delta-function sources under pseudosphere boundary conditions. We show that both the semiclassical wavefunction and the gravity solution are determined by a universal object, namely a classical Virasoro vacuum block on the sphere. In doing so we derive a version of Polyakov's conjecture, as well as an existence criterion, for classical Liouville theory on the pseudosphere. We also discuss how some of these results are modified when considering closed universes with compact spatial slices.

hep-th↗

arXiv2022

Ensemble Machine Learning Model Trained on a New Synthesized Dataset Generalizes Well for Stress Prediction Using Wearable Devices

Introduction. We investigate the generalization ability of models built on datasets containing a small number of subjects, recorded in single study protocols. Next, we propose and evaluate methods combining these datasets into a single, large dataset. Finally, we propose and evaluate the use of ensemble techniques by combining gradient boosting with an artificial neural network to measure predictive power on new, unseen data. Methods. Sensor biomarker data from six public datasets were utilized in this study. To test model generalization, we developed a gradient boosting model trained on one dataset (SWELL), and tested its predictive power on two datasets previously used in other studies (WESAD, NEURO). Next, we merged four small datasets, i.e. (SWELL, NEURO, WESAD, UBFC-Phys), to provide a combined total of 99 subjects,. In addition, we utilized random sampling combined with another dataset (EXAM) to build a larger training dataset consisting of 200 synthesized subjects,. Finally, we developed an ensemble model that combines our gradient boosting model with an artificial neural network, and tested it on two additional, unseen publicly available stress datasets (WESAD and Toadstool). Results. Our method delivers a robust stress measurement system capable of achieving 85% predictive accuracy on new, unseen validation data, achieving a 25% performance improvement over single models trained on small datasets. Conclusion. Models trained on small, single study protocol datasets do not generalize well for use on new, unseen data and lack statistical power. Ma-chine learning models trained on a dataset containing a larger number of varied study subjects capture physiological variance better, resulting in more robust stress detection.

cs.LG↗

arXiv2022

Generalizable machine learning for stress monitoring from wearable devices: A systematic literature review

Introduction. The stress response has both subjective, psychological and objectively measurable, biological components. Both of them can be expressed differently from person to person, complicating the development of a generic stress measurement model. This is further compounded by the lack of large, labeled datasets that can be utilized to build machine learning models for accurately detecting periods and levels of stress. The aim of this review is to provide an overview of the current state of stress detection and monitoring using wearable devices, and where applicable, machine learning techniques utilized. Methods. This study reviewed published works contributing and/or using datasets designed for detecting stress and their associated machine learning methods, with a systematic review and meta-analysis of those that utilized wearable sensor data as stress biomarkers. The electronic databases of Google Scholar, Crossref, DOAJ and PubMed were searched for relevant articles and a total of 24 articles were identified and included in the final analysis. The reviewed works were synthesized into three categories of publicly available stress datasets, machine learning, and future research directions. Results. A wide variety of study-specific test and measurement protocols were noted in the literature. A number of public datasets were identified that are labeled for stress detection. In addition, we discuss that previous works show shortcomings in areas such as their labeling protocols, lack of statistical power, validity of stress biomarkers, and generalization ability. Conclusion. Generalization of existing machine learning models still require further study, and research in this area will continue to provide improvements as newer and more substantial datasets become available for study.

cs.AI↗

arXiv2021

Conformal Blocks and Bilocal Vertex Operator Transition Amplitudes

We revisit the construction of the 2d conformal blocks of primary operator four-point functions as bilocal vertex operator correlators. We find an additional interpretation as a path integral over the reparametrizations of an intermediate cylinder. As a consequence we bridge the gap between the K\"ahler quantization of virasoro coadjoint orbits, $SL(2,\mathbb{R})$ Chern-Simons theory and the reparametrization formalism of 2d CFT that has made an appearance in recent literature.

hep-th↗

arXiv2020

The Large-$c$ Virasoro Identity Block is a Semi-Classical Liouville Correlator

It will be shown analytically that the light sector of the identity block of a mixed heavy-light correlator in the large central charge limit is given by a correlation function of light operators on an effective background geometry. This geometry is generated by the presence of the heavy operators. It is shown that this background geometry is a solution to the Liouville equation of motion sourced by corresponding heavy vertex operators and subsequently that the light sector of the identity block matches the Liouville correlation function in the semi-classical limit. This method effectively captures the spirit of Einstein gravity as a theory of dynamical geometry in AdS/CFT. The reason being that Liouville theory is closely related to semi-classical asymptotically AdS$_3$ gravity.

hep-th↗

arXiv2018

Vacuum block thermalization in semi-classical 2d CFT

The universal nature of black hole collapse in asymptotically $AdS_3$ gravitational theories suggests that its holographic dual process, thermalization, should similarly be fixed by the universal features of 2d CFT with large central charge $c$. It is known that non-equilibrium states with scaling dimensions of order $c$ can be sorted into states that eventually thermalize and those that fail to do so. By proving an equivalence between bounded Virasoro coadjoint orbits and certain (in)stability intervals of Hill's equation it is shown that a state that fails to thermalize can be promoted to a thermalizing state by preparing the system beforehand with an energy greater than an appropriate threshold energy. It is generally a difficult problem to ascertain whether a state will thermalize or not. As partial progress to this problem a set of lower bounds are presented for the treshold energy, which can alternatively be interpreted as criteria for thermalization.

hep-th↗

arXiv2018

On the universality of late-time correlators in semi-classical 2d CFTs

In the framework of AdS$_3$/ CFT$_2$ correspondence, we present a systematic analysis of the late time thermalization of a two dimensional CFT state created by insertion of small number of heavy operators on the vacuum. We show that at late Lorentzian time, the universal features of this thermalization are solely captured by the eigenvalues of the monodromy matrix corresponding to the solutions of the uniformization equation. We discuss two different ways to extract the monodromy eigenvalues while bypassing the need for finding explicitly the full monodromy matrix - first, using a monodromy preserving diffeomorphism and second using Chen-Simons formulation of gravity in AdS$_3$. Both of the methods yield the same precise relation between the eigenvalues and the final black hole temperature at late Lorentzian time.

hep-th↗

arXiv2016

Large-N correlation functions in ${\cal N} = 2$ superconformal QCD

We study extremal correlation functions of chiral primary operators in the large-N SU(N) ${\cal N} = 2$ superconformal QCD theory and present new results based on supersymmetric localization. We discuss extensively the basis-independent data that can be extracted from these correlators using the leading order large-N matrix model free energy given by the four-sphere partition function. Special emphasis is given to single-trace 2- and 3-point functions as well as a new class of observables that are scalars on the conformal manifold. These new observables are particular quadratic combinations of the structure constants of the chiral ring. At weak 't Hooft coupling we present perturbative results that, in principle, can be extended to arbitrarily high order. We obtain closed-form expressions up to the first subleading order. At strong coupling we provide analogous results based on an approximate Wiener-Hopf method.

hep-th↗

arXiv2015

Generalized Additivity in Unitary Conformal Field Theories

It was demonstrated in recent work that $d=4$ unitary CFT's satisfy a special property: if a scalar operator with conformal dimension $Δ$ exists in the operator spectrum, then the conformal bootstrap demands that large spin primary operators have to exist in the operator spectrum of the CFT with a conformal twist close to $2Δ+2N$ for any integer $N$. In this paper the conformal bootstrap methods that were used to find the anomalous dimension of the $N=0$ operators have been generalized to find the anomalous dimension of all large spin operators of this class. In AdS these operators can be interpreted as the excited states of the product states of objects that were found in other works.

hep-th↗