Search

Early identification of respiratory irregularities is critical for improving lung health and reducing global mortality rates. The analysis of respiratory sounds plays a significant role in characterizing the respiratory system's condition and identifying abnormalities. The main contribution of this study is to investigate the performance when the input data, represented by cochleogram, is used to feed the Vision Transformer architecture, since this input classifier combination is the first time it has been applied to adventitious sound classification to our knowledge. Although ViT has shown promising results in audio classification tasks by applying self attention to spectrogram patches, we extend this approach by applying the cochleogram, which captures specific spectro-temporal features of adventitious sounds. The proposed methodology is evaluated on the ICBHI dataset. We compare the classification performance of ViT with other state of the art CNN approaches using spectrogram, Mel frequency cepstral coefficients, constant Q transform, and cochleogram as input data. Our results confirm the superior classification performance combining cochleogram and ViT, highlighting the potential of ViT for reliable respiratory sound classification. This study contributes to the ongoing efforts in developing automatic intelligent techniques with the aim to significantly augment the speed and effectiveness of respiratory disease detection, thereby addressing a critical need in the medical field.

Training machine learning models based on neural networks requires large datasets, which may contain sensitive information. The models, however, should not expose private information from these datasets. Differentially private SGD [DP-SGD] requires the modification of the standard stochastic gradient descent [SGD] algorithm for training new models. In this short paper, a novel regularization strategy is proposed to achieve the same goal in a more efficient manner.

Densities, $\rho$, and kinematic viscosities, $\nu$, have been determined at atmospheric pressure and at (293.15-303.15) K for binary mixtures formed by methanol and one linear polyether of the type CH$_3$-O-(CH$_2$CH$_2$O)$_n$-CH$_3$ ($n$ = 2,3,4). The $\rho$ values are used to compute excess molar volumes, $V_{\text{m}}^{\text{E}}$, and, together with $\nu$ results, dynamic viscosities ($\eta$). Deviations from linear dependence on mole fraction for viscosity, $\Delta \eta$, are also provided. Different semi-empirical equations have been employed to correlate viscosity data. Particularly, the equations used are: Grunberg-Nissan, Hind, Frenkel, Katti-Chaudhri, McAllister and Heric. Calculations show that better results are obtained from the Hind equation. The $V_{\text{m}}^{\text{E}}$ values are large and negative and contrast with the positive excess molar enthalpies, $H_{\text{m}}^{\text{E}}$, available in the literature, for these systems. This indicates that structural effects are dominant. The $\Delta \eta$ results are positive and correlate well with the difference in volume of the mixture compounds confirming the importance of structural effects. The temperature dependences of $\eta$ and of the molar volume have been used to calculate enthalpies, entropies and Gibbs energies, $\Delta G^*$, of viscous flow. It is demonstrated that $\Delta G^*$ is essentially determined by enthalpic effects. Methanol + CH$_3$-O-(CH$_2$CH$_2$O)$_n$-CH$_3$ mixtures have been treated in the framework of the ERAS model. Results on $H_{\text{m}}^{\text{E}}$ are acceptable, while the composition dependence of the $V_{\text{m}}^{\text{E}}$ curves is poorly represented. This has been ascribed to the existence of strong dipolar and structural effects in the present solutions.

We present a novel metric designed, among other applications, to quantify biased behaviors of machine learning models. As its core, the metric consists of a new similarity metric between score distributions that balances both their general shapes and tails' probabilities. In that sense, our proposed metric may be useful in many application areas. Here we focus on and apply it to the operational evaluation of face recognition systems, with special attention to quantifying demographic biases; an application where our metric is especially useful. The topic of demographic bias and fairness in biometric recognition systems has gained major attention in recent years. The usage of these systems has spread in society, raising concerns about the extent to which these systems treat different population groups. A relevant step to prevent and mitigate demographic biases is first to detect and quantify them. Traditionally, two approaches have been studied to quantify differences between population groups in machine learning literature: 1) measuring differences in error rates, and 2) measuring differences in recognition score distributions. Our proposed Comprehensive Equity Index (CEI) trade-offs both approaches combining both errors from distribution tails and general distribution shapes. This new metric is well suited to real-world scenarios, as measured on NIST FRVT evaluations, involving high-performance systems and realistic face databases including a wide range of covariates and demographic groups. We first show the limitations of existing metrics to correctly assess the presence of biases in realistic setups and then propose our new metric to tackle these limitations. We tested the proposed metric with two state-of-the-art models and four widely used databases, showing its capacity to overcome the main flaws of previous bias metrics., Comment: Accepted paper for the 27th International Conference on Pattern Recognition (ICPR) 2024

The emergence of distributed generation and the electrification of demand have opened the possibility for prosumers to participate in electricity markets, receiving economic benefits on their bills and contributing to the reduction of carbon emissions, aligning with United Nations Sustainable Development Goal 7. Consumers and prosumers can participate through implicit and explicit demand flexibility and (collective) self-consumption. This study analyses the potential markets in which prosumers can participate and indicates whether these are currently open. The markets studied include day-ahead, intraday, ancillary services, adequacy services, constraint management, and local flexibility markets. Additionally, collective self-consumption is analysed as a service through which prosumers can participate in the electricity market. Previous studies are usually focused on a single market or in a single country, making impossible a complete comparison. This analysis has been done in Spain, Italy, Croatia, and the United Kingdom as representative countries to obtain a methodology to assess countries' openness to prosumer participation in electricity markets, comparing regulatory frameworks and assigning scores based on their prosumer inclusion across various markets. This work updates current literature reviews with the changes and a new description of local market designs in Spain. This methodology can be used to compare other countries' grade of openness. The results of this study show that the analysed countries can be categorised into three groups: almost open, partially open, and closed markets. Analysing the differences, recommendations on the following steps to foster user participation are suggested for each group.

Interpreting the observations of exoplanet atmospheres to constrain physical and chemical properties is typically done using Bayesian retrieval techniques. Because these methods require many model computations, a compromise is made between model complexity and run time. Reaching this compromise leads to the simplification of many physical and chemical processes (e.g. parameterised temperature structure). Here we implement and test sequential neural posterior estimation (SNPE), a machine learning inference algorithm, for exoplanet atmospheric retrievals. The goal is to speed up retrievals so they can be run with more computationally expensive atmospheric models, such as those computing the temperature structure using radiative transfer. We generate 100 synthetic observations using ARCiS (ARtful Modeling Code for exoplanet Science, an atmospheric modelling code with the flexibility to compute models in varying degrees of complexity) and perform retrievals on them to test the faithfulness of the SNPE posteriors. The faithfulness quantifies whether the posteriors contain the ground truth as often as we expect. We also generate a synthetic observation of a cool brown dwarf using the self-consistent capabilities of ARCiS and run a retrieval with self-consistent models to showcase the possibilities that SNPE opens. We find that SNPE provides faithful posteriors and is therefore a reliable tool for exoplanet atmospheric retrievals. We are able to run a self-consistent retrieval of a synthetic brown dwarf spectrum using only 50,000 forward model evaluations. We find that SNPE can speed up retrievals between $\sim2\times$ and $\geq10\times$ depending on the computational load of the forward model, the dimensionality of the observation, and the signal-to-noise ratio of the observation. We make the code publicly available for the community on Github., Comment: Accepted for publication at A&A

WASP-107b is a warm ($\sim$740 K) transiting planet with a Neptune-like mass of $\sim$30.5 $M_{\oplus}$ and Jupiter-like radius of $\sim$0.94 $R_{\rm J}$, whose extended atmosphere is eroding. Previous observations showed evidence for water vapour and a thick high-altitude condensate layer in WASP-107b's atmosphere. Recently, photochemically produced sulphur dioxide (SO$_2$) was detected in the atmosphere of a hot ($\sim$1,200 K) Saturn-mass planet from transmission spectroscopy near 4.05 $\mu$m, but for temperatures below $\sim$1,000 K sulphur is predicted to preferably form sulphur allotropes instead of SO$_2$. Here we report the 9$\sigma$-detection of two fundamental vibration bands of SO$_2$, at 7.35 $\mu$m and 8.69 $\mu$m, in the transmission spectrum of WASP-107b using the Mid-Infrared Instrument (MIRI) of the JWST. This discovery establishes WASP-107b as the second irradiated exoplanet with confirmed photochemistry, extending the temperature range of exoplanets exhibiting detected photochemistry from $\sim$1,200 K down to $\sim$740 K. Additionally, our spectral analysis reveals the presence of silicate clouds, which are strongly favoured ($\sim$7$\sigma$) over simpler cloud setups. Furthermore, water is detected ($\sim$12$\sigma$), but methane is not. These findings provide evidence of disequilibrium chemistry and indicate a dynamically active atmosphere with a super-solar metallicity.

Brown dwarfs serve as ideal laboratories for studying the atmospheres of giant exoplanets on wide orbits as the governing physical and chemical processes in them are nearly identical. Understanding the formation of gas giant planets is challenging, often involving the endeavour to link atmospheric abundance ratios, such as the carbon-to-oxygen (C/O) ratio, to formation scenarios. However, the complexity of planet formation requires additional tracers, as the unambiguous interpretation of the measured C/O ratio is fraught with complexity. Isotope ratios, such as deuterium-to-hydrogen and 14N/15N, offer a promising avenue to gain further insight into this formation process, mirroring their utility within the solar system. For exoplanets only a handful of constraints on 12C/13C exist, pointing to the accretion of 13C-rich ice from beyond the disks' CO iceline. Here we report on the mid-infrared detection of the 14NH3 and 15NH3 isotopologues in the atmosphere of a cool brown dwarf with an effective temperature of 380 K in a spectrum taken with the Mid-InfraRed Instrument of the James Webb Space Telescope. As expected, our results reveal a 14N/15N value consistent with star-like formation by gravitational collapse, demonstrating that this ratio can be accurately constrained. Since young stars and their planets should be more strongly enriched in the 15N isotope, we expect that 15NH3 will be detectable in a number of cold, wide-separation exoplanets., Comment: Accepted by Nature. 28 pages, 7 figures, uses nature3.cls

Presentation Attack Detection (PAD) is a crucial stage in facial recognition systems to avoid leakage of personal information or spoofing of identity to entities. Recently, pulse detection based on remote photoplethysmography (rPPG) has been shown to be effective in face presentation attack detection. This work presents three different approaches to the presentation attack detection based on rPPG: (i) The physiological domain, a domain using rPPG-based models, (ii) the Deepfakes domain, a domain where models were retrained from the physiological domain to specific Deepfakes detection tasks; and (iii) a new Presentation Attack domain was trained by applying transfer learning from the two previous domains to improve the capability to differentiate between bona-fides and attacks. The results show the efficiency of the rPPG-based models for presentation attack detection, evidencing a 21.70% decrease in average classification error rate (ACER) (from 41.03% to 19.32%) when the presentation attack domain is compared to the physiological and Deepfakes domains. Our experiments highlight the efficiency of transfer learning in rPPG-based models and perform well in presentation attack detection in instruments that do not allow copying of this physiological feature., Comment: Preprint of the paper presented to the Workshop on IEEE 47th Annual Computers, Software, and Applications Conference (COMPSAC, 2023)

Sparsity is a growing trend in modern DNN models. Existing Sparse-Sparse Matrix Multiplication (SpMSpM) accelerators are tailored to a particular SpMSpM dataflow (i.e., Inner Product, Outer Product or Gustavsons), that determines their overall efficiency. We demonstrate that this static decision inherently results in a suboptimal dynamic solution. This is because different SpMSpM kernels show varying features (i.e., dimensions, sparsity pattern, sparsity degree), which makes each dataflow better suited to different data sets. In this work we present Flexagon, the first SpMSpM reconfigurable accelerator that is capable of performing SpMSpM computation by using the particular dataflow that best matches each case. Flexagon accelerator is based on a novel Merger-Reduction Network (MRN) that unifies the concept of reducing and merging in the same substrate, increasing efficiency. Additionally, Flexagon also includes a 3-tier memory hierarchy, specifically tailored to the different access characteristics of the input and output compressed matrices. Using detailed cycle-level simulation of contemporary DNN models from a variety of application domains, we show that Flexagon achieves average performance benefits of 4.59x, 1.71x, and 1.35x with respect to the state-of-the-art SIGMA-like, Sparch-like and GAMMA-like accelerators (265% , 67% and 18%, respectively, in terms of average performance/area efficiency)., Comment: To appear on ASPLOS 2023

Machine learning models work better when curated features are provided to them. Feature engineering methods have been usually used as a preprocessing step to obtain or build a proper feature set. In late years, autoencoders (a specific type of symmetrical neural network) have been widely used to perform representation learning, proving their competitiveness against classical feature engineering algorithms. The main obstacle in the use of autoencoders is finding a good architecture, a process that most experts confront manually. An automated autoencoder architecture search procedure, based on evolutionary methods, is proposed in this paper. The methodology is tested against nine heterogeneous data sets. The obtained results show the ability of this approach to find better architectures, able to concentrate most of the useful information in a minimized coding, in a reduced time., Comment: Paper submited to Integrated Computer-Aided Engineering