Search

Studying the excited states of doublets is challenging for their typically multiconfigurational character. We employ light-scanning-tunneling microscopy (light-STM) to investigate photon-induced currents on a single open-shell PTCDA anion molecule placed into a plasmonic nanocavity between a tip and a substrate, irradiated by laser. Submolecular mapping reveals a zero-bias bidirectional photocurrent strongly varying with the lateral position of the tip apex above the molecule. We elucidate the mechanism in terms of a theoretical model in which a multiconfigurational doublet state is excited and decays back to the anion ground state through sequential electron transfers with the tip and the substrate. The correspondence of the experimental and theoretical contrast proves the correlated character of the excited state which can be described as a superposition of two dominating electronic configurations. By applying bipolar voltage on the junction with the molecule, we switch the dominant recombination pathway from one of the configurations to the other, effectively disentangling the multiconfigurational state individual components through visualization of their Dyson orbitals, as corroborated by theoretical modelling., Comment: Manuscript + Supplementary information

Abdominal computed tomography (CT) scans are frequently performed in clinical settings. Opportunistic CT involves repurposing routine CT images to extract diagnostic information and is an emerging tool for detecting underdiagnosed conditions such as sarcopenia, hepatic steatosis, and ascites. This study utilizes deep learning methods to promote accurate diagnosis and clinical documentation. We analyze 2,674 inpatient CT scans to identify discrepancies between imaging phenotypes (characteristics derived from opportunistic CT scans) and their corresponding documentation in radiology reports and ICD coding. Through our analysis, we find that only 0.5%, 3.2%, and 30.7% of scans diagnosed with sarcopenia, hepatic steatosis, and ascites (respectively) through either opportunistic imaging or radiology reports were ICD-coded. Our findings demonstrate opportunistic CT's potential to enhance diagnostic precision and accuracy of risk adjustment models, offering advancements in precision medicine.

Ontologies are a standard tool for creating semantic schemata in many knowledge intensive domains of human interest. They are becoming increasingly important also in the areas that have been until very recently dominated by subsymbolic knowledge representation and machine-learning (ML) based data processing. One such area is information security, and specifically, malware detection. We thus propose PE Malware Ontology that offers a reusable semantic schema for Portable Executable (PE - the Windows binary format) malware files. This ontology is inspired by the structure of the EMBER dataset, which focuses on the static malware analysis of PE files. With this proposal, we hope to provide a unified semantic representation for the existing and future PE-malware datasets and facilitate the application of symbolic, neuro-symbolic, or otherwise explainable approaches in the PE-malware-detection domain, which may produce interpretable results described by the terms defined in our ontology. In addition, we also publish semantically treated EMBER data, including fractional datasets, to support the reproducibility of experiments on EMBER. We supplement our work with a preliminary case study, conducted using concept learning, to show the general feasibility of our approach. While we were not able to match the precision of the state-of-the-art ML tools, the learned malware discriminators were interesting and highly interpretable., Comment: arXiv admin note: substantial text overlap with arXiv:2301.00153

This study details our approach for the CASE 2024 Shared Task on Climate Activism Stance and Hate Event Detection, focusing on Hate Speech Detection, Hate Speech Target Identification, and Stance Detection as classification challenges. We explored the capability of Large Language Models (LLMs), particularly GPT-4, in zero- or few-shot settings enhanced by retrieval augmentation and re-ranking for Tweet classification. Our goal was to determine if LLMs could match or surpass traditional methods in this context. We conducted an ablation study with LLaMA for comparison, and our results indicate that our models significantly outperformed the baselines, securing second place in the Target Detection task. The code for our submission is available at https://github.com/NaiveNeuron/bryndza-case-2024, Comment: Accepted to the 7th Workshop on Challenges and Applications of Automated Extraction of Socio-political Events from Text (CASE 2024)

Single-molecule tip-enhanced Raman spectroscopy (TERS) under ultra-high vacuum (UHV) and cryogenic conditions enables exploration of the relations between the adsorption geometry, electronic state, and vibrational fingerprints of individual molecules. TERS capability of reflecting spin states in open-shell molecular configurations is yet unexplored. Here we use the tip of a scanning probe microscope to lift a perylene-3,4,9,10-tetracarboxylic dianhydride (PTCDA) molecule from a metal surface to bring it into an open-shell spin one-half anionic state. We reveal a correlation between the appearance of a Kondo resonance in the differential conductance spectroscopy and concurrent characteristic changes captured by the TERS measurements. Through a detailed investigation of various adsorbed and tip-contacted PTCDA scenarios, we infer that the Raman scattering on the suspended PTCDA is resonant with a higher excited state. Theoretical simulation of the vibrational spectra enables a precise assignment of the individual TERS peaks to high-symmetry Ag modes, including the fingerprints of the observed spin state. These findings highlight the potential of TERS in capturing complex interactions between charge, spin, and photophysical properties in nanoscale molecular systems, and suggest a pathway for designing spin-optical devices using organic molecules.

The spontaneous fluorescence rates of single-molecule emitters are typically on the order of nanoseconds. However coupling them with plasmonic nanostructures can substantially increase their fluorescence yields. The confinement between the tip and sample of a scanning tunneling microscope creates a tunable nanocavity, an ideal platform for exploring the yields and excitation decay rates of single-molecule emitters depending on the coupling strength to the nanocavity. With this setup we estimate the excitation lifetimes from the direct time-resolved measurements of the fluorescence decays of phthalocyanine adsorbates, decoupled from the metal substrates by ultrathin NaCl layers. It is found that nanosecond-range lifetimes prevail for the emitters away from the nanocavity, whereas for the tip approached to a molecule, we find a substantial effect of the nanocavity coupling, which reduces the lifetimes to a few picoseconds. An analysis is performed to investigate the crossover between the far-field and tip-enhanced photoluminescence regimes. This approach overcomes the drawbacks associated with the estimation of lifetimes for single molecules from their respective emission linewidths.

Hematite ${\alpha}-Fe_{2}O_{3}(0001)$ is the most-investigated iron oxide model system in photo and electrocatalytic research. The rich chemistry of Fe and O allows for many bulk and surface transformations, but their control is challenging. This has led to controversies regarding the structure of the topmost layers. This comprehensive study combines surface methods (nc-AFM, STM, LEED, and XPS) complemented by structural and chemical analysis of the near-surface bulk (HRTEM and EELS). The results show that a compact 2D layer constitutes the topmost surface of ${\alpha}-Fe_{2}O_{3}(0001)$; it is locally corrugated due to the mismatch with the bulk. Assessing the influence of naturally-occurring impurities shows that these can force the formation of surface phases that are not stable on pure samples. Impurities can also cause the formation of ill-defined inclusions in the subsurface and modify the oxidation phase diagram of hematite. The results provide a significant step forward in determining the hematite surface structure that is crucial for accurately modeling catalytic reactions. Combining surface and cross-sectional imaging provided the full view that is essential for understanding the evolution of the near-surface region of oxide surfaces under oxidative conditions.

Ontologies are a standard for semantic schemata in many knowledge-intensive domains of human interest. They are now becoming increasingly important also in areas until very recently dominated by subsymbolic representations and machine-learning-based data processing. One such area is information security, and more specifically malware detection. We propose PE Malware Ontology that offers a reusable semantic schema for Portable Executable (PE, Windows binary format) malware files. The ontology was inspired by the structure of the data in the EMBER dataset and it currently covers the data intended for static malware analysis. With this proposal, we hope to achieve: a) a unified semantic representation for PE malware datasets that are available or will be published in the future; (b) applicability of symbolic, neural-symbolic, or otherwise explainable approaches in the PE Malware domain that may lead to improved interpretability of results which may now be characterized by the terms defined in the ontology; and (c)by joint publishing of semantically treated EMBER data, including fractional datasets, also improved reproducibility of experiments.

Emergent quantum phenomena in two-dimensional van der Waal (vdW) magnets are largely governed by the interplay between the exchange and Coulomb interactions. The ability to tune the Coulomb interaction in such strongly correlated materials enables the precise control of spin-correlated flat-band states, bandgap (Eg) and unconventional magnetism, all of which are crucial for next-generation spintronics and magnonics applications. Here, we demonstrate a giant gate-tunable renormalization of spin-correlated flat-band states and bandgap in magnetic chromium tribromide (CrBr3) monolayers grown on graphene. Our gate-dependent scanning tunneling spectroscopy (STS) studies reveal that the inter-flat-band spacing and bandgap of CrBr3 can be continuously tuned by 120 meV and 240 meV respectively via electrostatic injection of carriers into the hybrid CrBr3/graphene system, equivalent to the modulation of the Cr on-site Coulomb repulsion energy by 500 meV. This can be attributed to the self-screening of CrBr3 arising from the gate-induced carriers injected into CrBr3, which dominates over the opposite trend from the remote screening of the graphene substrate. Precise tuning of the spin-correlated flat-band states and bandgap in 2D magnets via electrostatic modulation of Coulomb interactions not only provides new strategies for optimizing the spin transport channels but also may exert a crucial influence on the exchange energy and spin-wave gap, which could raise the critical temperature for magnetic order., Comment: 21 pages, 4 figures

The paper presents a method for spoken term detection based on the Transformer architecture. We propose the encoder-encoder architecture employing two BERT-like encoders with additional modifications, including convolutional and upsampling layers, attention masking, and shared parameters. The encoders project a recognized hypothesis and a searched term into a shared embedding space, where the score of the putative hit is computed using the calibrated dot product. In the experiments, we used the Wav2Vec 2.0 speech recognizer, and the proposed system outperformed a baseline method based on deep LSTMs on the English and Czech STD datasets based on USC Shoah Foundation Visual History Archive (MALACH)., Comment: Submitted to ICASSP 2023

The paper describes a novel approach to Spoken Term Detection (STD) in large spoken archives using deep LSTM networks. The work is based on the previous approach of using Siamese neural networks for STD and naturally extends it to directly localize a spoken term and estimate its relevance score. The phoneme confusion network generated by a phoneme recognizer is processed by the deep LSTM network which projects each segment of the confusion network into an embedding space. The searched term is projected into the same embedding space using another deep LSTM network. The relevance score is then computed using a simple dot-product in the embedding space and calibrated using a sigmoid function to predict the probability of occurrence. The location of the searched term is then estimated from the sequence of output probabilities. The deep LSTM networks are trained in a self-supervised manner from paired recognition hypotheses on word and phoneme levels. The method is experimentally evaluated on MALACH data in English and Czech languages.

In recent years, the standard hybrid DNN-HMM speech recognizers are outperformed by the end-to-end speech recognition systems. One of the very promising approaches is the grapheme Wav2Vec 2.0 model, which uses the self-supervised pretraining approach combined with transfer learning of the fine-tuned speech recognizer. Since it lacks the pronunciation vocabulary and language model, the approach is suitable for tasks where obtaining such models is not easy or almost impossible. In this paper, we use the Wav2Vec speech recognizer in the task of spoken term detection over a large set of spoken documents. The method employs a deep LSTM network which maps the recognized hypothesis and the searched term into a shared pronunciation embedding space in which the term occurrences and the assigned scores are easily computed. The paper describes a bootstrapping approach that allows the transfer of the knowledge contained in traditional pronunciation vocabulary of DNN-HMM hybrid ASR into the context of grapheme-based Wav2Vec. The proposed method outperforms the previously published system based on the combination of the DNN-HMM hybrid ASR and phoneme recognizer by a large margin on the MALACH data in both English and Czech languages.

Recently, the performance of blind speech separation (BSS) and target speech extraction (TSE) has greatly progressed. Most works, however, focus on relatively well-controlled conditions using, e.g., read speech. The performance may degrade in more realistic situations. One of the factors causing such degradation may be intrinsic speaker variability, such as emotions, occurring commonly in realistic speech. In this paper, we investigate the influence of emotions on TSE and BSS. We create a new test dataset of emotional mixtures for the evaluation of TSE and BSS. This dataset combines LibriSpeech and Ryerson Audio-Visual Database of Emotional Speech and Song (RAVDESS). Through controlled experiments, we can analyze the impact of different emotions on the performance of BSS and TSE. We observe that BSS is relatively robust to emotions, while TSE, which requires identifying and extracting the speech of a target speaker, is much more sensitive to emotions. On comparative speaker verification experiments we show that identifying the target speaker may be particularly challenging when dealing with emotional speech. Using our findings, we outline potential future directions that could improve the robustness of BSS and TSE systems toward emotional speech., Comment: Accepted to IWAENC 2022

In this paper, we present our progress in pretraining Czech monolingual audio transformers from a large dataset containing more than 80 thousand hours of unlabeled speech, and subsequently fine-tuning the model on automatic speech recognition tasks using a combination of in-domain data and almost 6 thousand hours of out-of-domain transcribed speech. We are presenting a large palette of experiments with various fine-tuning setups evaluated on two public datasets (CommonVoice and VoxPopuli) and one extremely challenging dataset from the MALACH project. Our results show that monolingual Wav2Vec 2.0 models are robust ASR systems, which can take advantage of large labeled and unlabeled datasets and successfully compete with state-of-the-art LVCSR systems. Moreover, Wav2Vec models proved to be good zero-shot learners when no training data are available for the target ASR task., Comment: to be published in Proceedings of INTERSPEECH 2022

Interplay between motion of nuclei and excited electrons in molecules plays a key role both in biological and artificial nanomachines. Here we provide a detailed analysis of coupling between quantized librational modes (librons) and charged excited states (trions) on single phthalocyanine dyes adsorbed on a surface. By means of tunnelling electron-induced electroluminescence, we identify libronic progressions on a {\mu}eV energy range in spectra of chirally adsorbed phthalocyanines, which are otherwise absent from spectra of symmetrically adsorbed species. Experimentally measured libronic spectra match very well the theoretically calculated libron eigenenergies and peak intensities (Franck-Condon factors) and reveal an unexpected depopulation channel for the zero libron of the excited state that can be effectively controlled by tuning the size of the nanocavity. Our results showcase the possibility of characterizing the dynamics of molecules by their low-energy molecular modes using {\mu}eV-resolved tip-enhanced spectroscopy.

Scanning tunneling and atomic force microscopies (STM/nc-AFM) are rapidly progressing to offer unprecedented spatial resolution of a diverse array of chemical species. In particular, they are employed to characterize on-surface chemical reactions by directly examining precursors and products. Chiral effects and self-assembled structures can also be investigated. This open source, modular, python based scheme automates the categorization of a variety of molecules present in medium sized (10$\times$10 to 100$\times$100 nm) scanned probe images.

One-dimensional metal-organic chains often possess a complex magnetic structure susceptible to be modified by a alteration of their chemical composition. The possibility to tune their magnetic properties provides an interesting playground to explore quasiparticle interactions in low-dimensional systems. Despite the great effort invested so far, a detailed understanding of the interactions governing the electronic and magnetic properties of the low-dimensional systems is still incomplete. One of the reasons is the limited ability to characterize their magnetic properties at the atomic scale. Here, we provide a comprehensive study of the magnetic properties of metal-organic one-dimensional (1D) coordination polymers consisting of 2,5-diamino-1,4-benzoquinonediimine ligands coordinated with Co or Cr atoms synthesized in ultra-high vacuum conditions on a Au(111) surface. A combination of an integral X-ray spectroscopy with local-probe inelastic electron tunneling spectroscopy corroborated by multiplet analysis, density functional theory, and inelastic electron tunneling simulations enable us to obtain essential information about their magnetic structure, including the spin magnitude and orientation at the magnetic atoms, as well as the magnetic anisotropy., Comment: 35 pages, 8 Figures, 3 tables

In typical multi-talker speech recognition systems, a neural network-based acoustic model predicts senone state posteriors for each speaker. These are later used by a single-talker decoder which is applied on each speaker-specific output stream separately. In this work, we argue that such a scheme is sub-optimal and propose a principled solution that decodes all speakers jointly. We modify the acoustic model to predict joint state posteriors for all speakers, enabling the network to express uncertainty about the attribution of parts of the speech signal to the speakers. We employ a joint decoder that can make use of this uncertainty together with higher-level language information. For this, we revisit decoding algorithms used in factorial generative models in early multi-talker speech recognition systems. In contrast with these early works, we replace the GMM acoustic model with DNN, which provides greater modeling power and simplifies part of the inference. We demonstrate the advantage of joint decoding in proof of concept experiments on a mixed-TIDIGITS dataset., Comment: submitted to Interspeech 2022

Entanglement of excitons holds great promise for the future of quantum computing, which would use individual molecular dyes as building blocks of their circuitry. Even though entangled excitonic eigenstates emerging in coupled molecular assemblies can be detected by far-field spectroscopies, access to the individual modes in real space will bring the much needed insight into the photophysics of these fascinating quantum phenomena. Here we combine tip-enhanced spectromicroscopy with atomic force microscopy to inspect delocalized single-exciton states of charged molecular assemblies engineered from individual perylenetetracarboxylic dianhydride molecules. Hyperspectral mapping of the eigenstates and comparison with calculated many-body optical transitions reveals a second low-lying excited state of the anion monomers and its role in the exciton entanglement within the assemblies. We also demonstrate control over the coupling by switching the assembly charge states. Our results reveal the possibility of tailoring excitonic properties of organic dye aggregates for advanced functionalities and establish the methodology to address them individually at the nanoscale.

In this paper, we present our progress in pre-training monolingual Transformers for Czech and contribute to the research community by releasing our models for public. The need for such models emerged from our effort to employ Transformers in our language-specific tasks, but we found the performance of the published multilingual models to be very limited. Since the multilingual models are usually pre-trained from 100+ languages, most of low-resourced languages (including Czech) are under-represented in these models. At the same time, there is a huge amount of monolingual training data available in web archives like Common Crawl. We have pre-trained and publicly released two monolingual Czech Transformers and compared them with relevant public models, trained (at least partially) for Czech. The paper presents the Transformers pre-training procedure as well as a comparison of pre-trained models on text classification task from various domains., Comment: https://huggingface.co/fav-kky

In this study we demonstrate the viability of deploying BERT-style models to serverless environments in a production setting. Since the freely available pre-trained models are too large to be deployed in this way, we utilize knowledge distillation and fine-tune the models on proprietary datasets for two real-world tasks: sentiment analysis and semantic textual similarity. As a result, we obtain models that are tuned for a specific domain and deployable in serverless environments. The subsequent performance analysis shows that this solution results in latency levels acceptable for production use and that it is also a cost-effective approach for small-to-medium size deployments of BERT models, all without any infrastructure overhead., Comment: NAACL-HLT 2021 Industry Track Camera Ready

Charged optical excitations (trions) generated by charge carrier injection are crucial for emerging optoelectronic technologies as they can be produced and manipulated by electric fields. Trions and neutral excitons can be efficiently induced in single molecules by means of tip-enhanced spectromicroscopic techniques. However, little is known of the exciton-trion dynamics at single molecule level as this requires methods permitting simultaneous sub-nanometer and sub-nanosecond characterization. Here, we investigate exciton-trion dynamics by phase fluorometry, combining radio-frequency modulated scanning tunnelling luminescence with time-resolved single photon detection. We generate excitons and trions in single Zinc Phthalocyanine (ZnPc) molecules on NaCl/Ag(111), determine their dynamics and trace the evolution of the system in the picosecond range with atomic resolution. In addition, we explore dependence of effective lifetimes on bias voltage and propose a conversion of neutral excitons into trions via charge capture as the primary mechanism of trion formation.

In hydrogen bonded systems, nuclear quantum effects such as zero-point motion and tunneling can significantly affect their material properties through underlying physical and chemical processes. Presently, direct observation of the influence of nuclear quantum effects on the strength of hydrogen bonds with resulting structural and electronic implications remains elusive, leaving opportunities for deeper understanding to harness their fascinating properties. We studied hydrogen-bonded one-dimensional quinonediimine molecular networks which may adopt two isomeric electronic configurations via proton transfer. Herein, we demonstrate that concerted proton transfer promotes a delocalization of {\pi}-electrons along the molecular chain, which enhances the cohesive energy between molecular units, increasing the mechanical stability of the chain and giving rise to new electronic in-gap states localized at the ends. These findings demonstrate the identification of a new class of isomeric hydrogen bonded molecular systems where nuclear quantum effects play a dominant role in establishing their chemical and physical properties. We anticipate that this work will open new research directions towards the control of mechanical and electronic properties of low-dimensional molecular materials via concerted proton tunneling.

We investigate electroluminescence of single molecular emitters on NaCl on Ag(111) and Au(111) with submolecular resolution in a low-temperature scanning probe microscope with tunneling current, atomic force and light detection capabilities. Role of the tip state is studied in the photon maps of a prototypical emitter, zinc phthalocyanine (ZnPc), using metal and CO-metal tips. CO-functionalization is found to have a dramatic impact on the resolution and contrast of the photon maps due to the localized overlap of the p-orbitals on the tip with the molecular orbitals of the emitter. The possibility of using the same CO-functionalized tip for tip-enhanced photon detection and high resolution atomic force is demonstrated. We study the electroluminescence of ZnPc, induced by charge carrier injection at sufficiently high bias voltages. We propose that the distinct level alignment of the ZnPc frontier orbital with the Au(111) and Ag(111) Fermi levels governs the primary excitation mechanisms as the injection of electrons and holes from the tip into the molecule, respectively. These findings put forward the importance of the tip status in the photon maps and contribute to a better understanding of the photophysics of organic molecules on surfaces., Comment: 15 pages, 5 figures

Iron oxides are among the most abundant compounds on Earth and have consequently been studied and used extensively in industrial processes. Despite these efforts, concrete understanding of some of their surface phase structures has remained elusive, in particular the oxidized ${\alpha}-Fe_2O_3(0001)$ hematite surface. We detail an optimized recipe to produce this phase over the entire hematite surface and study the geometrical parameters and composition of its complex structure by means of atomically resolved microscopy, electron diffraction and surface-sensitive spectroscopies. We conclude that the oxidized ${\alpha}-Fe_2O_3(0001)$ surface is terminated by a two-dimensional iron oxide with structure, lattice parameters, and orientation different from the bulk substrate. Using total-energy density functional theory for simulation of a large-scale atomic model, we identify the structure of the surface layer as antiferromagnetic, conductive $1T-FeO_2$ attached on half-metal terminated bulk. The model succeeds in reproducing the characteristic modulations observed in the atomically resolved images and electron diffraction patterns.

Determination of the molecular Kondo temperature $T_K$ poses a challenge in most cases when the experimental temperature cannot be tuned to a sufficient extent. We show how this ambiguity can be resolved if additional control parameters are present, such as magnetic field and mechanical gating. We record the evolution of the differential conductance by lifting an individual molecule from the metal surface with the tip of a scanning tunneling microscope. By fitting the measured conductance spectra with the single impurity Anderson model we are able to demonstrate that the lifting tunes the junction continuously from the strongly correlated Kondo-singlet to the free spin $1/2$ ground state. In the crossover regime, where $T_K$ is similar to the temperature of experiment, the fitting yields ambiguous estimates of $T_K$ varying by an order of magnitude. We show that analysis of the conductance measured in two distinct external magnetic fields can be used to resolve this problem., Comment: Main text: 10 pages, 6 figures; Supplementary Information: 11 pages, 5 figures

We construct a model of the Cyclic Universe from a joint theory of General relativity, Thermodynamics and Quantum information theory. Friedmann equations and the thermodynamical Gibbs-Duhem relation determine a general form of the Hubble function which predicts a dynamical Dark Energy (DE) and a dynamical Dark Matter (DM) described by new entropic terms and by the equations of state w_0=-1 and w_M=0, respectively, at all z. We posit that the spacetime has a quantum structure described by the Quantum information theory. We identify the space quanta \rho with two-qubit quantum states of of massless gravitons with helicity states |\pm 2>. All space quanta carry quantum information entropy S(\rho). All entangled quanta carry entanglement entropy S_E(\rho) and form DE. All non-entangled quanta form DM. In the absence of Baryonic matter DM and SE are described by probability distributions p(\vec{x},S) and q(\vec{x},\chi) where \chi=S_E(\rho)+S(\rho). Fisher information metric generates from these distributions the vacuum gravitational fields of DM and DE. Tn the presence of the Baryonic matter the distributions are displaced p->p' and q->q'. Fisher metric then defines the displaced fields. In Einstein's theory of General relativity Space is a gravitational field which we identify with the gravitational fields of DE and DM. The theory predicts the existence of a new "residual" matter term with equation of state w_r=-1/3 in the Hubble function, and a negative spatial curvature the consequence of which are constraints on cosmological parameters. We recover Robertson-Walker metric and the Friedmann equations from the gravitational fields of Dark Energy and Dark Matter at cosmological scales. The predictions of the theory are tested and confirmed in the Part II of this work., Comment: 44 pages including 2 appendices; equations (3.17) and (4.29) corrected, previous equations (3.35) and (3.36) deleted, equations (9.20) and (9.21) added; Subsection IV.D added; change in the address