This paper presents a monolithic, cost-effective, label free, low power, robust 3x2 fully differential ISFET pixel array aimed at pH sensing. Each pixel is based on a differential readout architecture which eliminates need for additional non-ideality compensation circuitry. Additionally, the design offers the capability to correct for residual drift and trapped charge offset through substrate-based voltage adjustment during calibration. The design also provides a unique ability for selecting pH sensitivity according to the desired mode of operation which has not been demonstrated previously. We present experimental results obtained from the pixel array fabricated in an unmodified AMS 0.35μm CMOS process using a 3.3V power supply. Each pixel demonstrates a differential sensitivity of 61mV/pH when operated in high sensitivity mode and 25mV/pH when operated in high non-ideality rejection mode over the entire pH range, with the ability to remove the already small drift of 0.3 mV/min and any offset mismatch by adjusting the substrate voltage

Deep Learning (DL) has become a popular method in various industries to support human decision-making in areas such as gaming, news, finance, biology, medicine, and engineering. Although statistical learning has been used in clinical decision support for some time, deep learning has not been widely implemented in the clinical field, apart from image-related tasks. Clinical natural language processing (NLP) in decision support systems continues to rely mostly on rule-based or statistical solutions, unlike other domains that have adopted deep learning. The development of clinical NLP applications in languages other than English, such as Japanese, is still lagging behind. This paper offers a review on the challenges that deep learning faces in clinical NLP and explores potentials and opportunities in the field. The purpose is to provide a broad understanding of the clinical NLP landscape for deep learning researchers and inspire them to get involved in this field.

In this letter, we provide new insights into the mutual coupling between antennas, using spatial distribution of near-field reactive Helicity and period-averaged EM Lagrangian density. First, we analytically show that reactive Helicity (given by the real part of complex Helicity) in frequency domain always becomes zero in the far-field region of generic linearly and circularly polarized antennas. Furthermore, we introduce the concept of ‘instantaneous Helicity’ (analogous to instantaneous Poynting vector) to calculate reactive Helicity for antennas excited by arbitrary time-domain signals (eg. short-duration pulses). We analytically demonstrate the fundamental connection between this instantaneous Helicity, EM inertia density and EM Lagrangian density. Finally, using MoM-based MATLAB Antenna Toolbox, we compute the distribution (contour-plots) of normalized reactive Helicity and period-averaged Lagrangian density around a two-port system of thin-wire dipoles in frequency domain, revealing complex near-field inter-element interactions. Such visualization of mutual coupling via reactive Helicity and period-averaged EM Lagrangian density can enable antenna array designers to go beyond the traditional S-parameter approach, and identify regions containing near-zone non-TEM fields and non-propagating energy density, which may contribute constructively/destructively to the port-to-port mutual coupling.

The software industry employs few women and underrepresented minorities. Several initiatives from organizations, practitioners, and OSS communities address equity, diversity, and inclusion in software development. However, most of these initiatives deal in a generalist and simplistic way, often confusing diversity with gender diversity, reinforcing an industry with a homogeneous workforce, which has little room for the inclusion of sub represented communities. One of these communities is the transgender, who suffers from several intersections of oppression concerning gender, race, class, age, and disabilities, thus having voices erased in the industry. Our research aims to identify the concepts of diversity in Software Engineering (SE) from a transgender perspective. We conduct a Systematic Grey Literature Review (GLR)  to map the concepts around diversity in SE, focused on trans perspective. We detail our GLR, and inspired in Grounded Theory (GT) we have a coding protocol to build a concept map. As a result, we obtained a concept map format, which illustrate some of the complexity of diversity issues in SE focused on trans people and how organizations and developers can contribute to trans inclusion in the software industry. The paper starts a discussion regarding the perspective of trans people in the area.  Research in diversity should observe the oppressions that exist and how they interact without simplifying and isolating issues such as treating gender in a binary way and not interconnected with other issues such as race. SE industry is strategic to the trans community. Although communities declare themselves to be diverse and open to entry, concrete actions are lacking.

This preprint collection concerns the proof-of-principle study of an adaptive algorithm used in a long-term sensory home training. It consists of the description of the adaptive procedure and supplementary material to characterize the algorithm. The main aim of the study was to determine whether the designed adaptive framework based on operant learning and brain-plasticity principles is able to promote learning over a long period of time, and whether indicators of learning can be derived from individual training data. The increase of task demands over time was shown in combination with positive learning effects at individual level. The study represents the proof-of-principle of the methods which can form the basis to design training procedures engaging simple and complex abilities such as sensory perception, sensory integration, motor control, and testing of complex psychometric functions.

This study analyzes optical code division multiplexed access system coding with two different types of spectral amplitude coding (SAC): Double Weight (DW) codes and Multiple Diagonal (MD) codes. These two encoding types differ in code structure and channel bandwidth length (the number of wavelengths representing the channel’s bandwidth), as well as Q-Factor, BER, and MAI values. This study analyzes systems with different number of users over single-mode optical fiber and presents a comparison of both. DW codes have a shorter length code than MD codes, and although this technique can generate unique code words for each user, it will break the system using MD encoding. This is another reason to increase bandwidth for MD codes.

Characteristics of high sensitivity MEMS pressure sensor chip for 10 kPa utilizing a novel electrical circuit are presented. The electrical circuit uses piezosensitive differential amplifier with negative feedback loop (PDA-NFL) based on two bipolar-junction transistors (BJT). The BJT has a vertical structure of n-p-n type (V-NPN) formed on a non-deformable chip area. The circuit contains eight piezoresistors located on a profiled membrane in the areas of maximum mechanical stresses. The experimental results prove that pressure sensor chip PDA-NFL with 4.0×4.0 mm2 chip area has sensitivity S = 10.1 ± 2.3 mV/V/kPa with nonlinearity of 2KNL = 0.26 ± 0.12 %/FS (pressure is applied from the back side of pressure sensor chip).

This paper proposes a modified natural gradient descent, containing fractional derivatives of Riemann-Liouville, Caputo and Grunwald-Letnikov types. Such approach belongs to information-geometric optimization methods, which take into account not only directions of gradients or momentum parameters, but the convexity (curvature) of minimizing function. This technique, comparing with second order optimization algorithms, lets to increase the rate of convergence. With fractional order derivatives it is possible to adjust the descent toward the neighborhood of the global minimum. In experiments, we demonstrated the increasing accuracy of image recognition on MNIST and CIFAR10, using the proposed optimization algorithm.

The Covid-19 Pandemic has renewed interest in contactless vital signs monitoring using state-of-the-art computer vision, which can efficiently screen for symptoms while reducing the risk of disease transmission. Despite the promising perfor- mance, the use of static camera setups requires subjects to remain static inside a field of view (FoV) for a pre-specified duration. Due to inconsistent ambient environmental conditions, the transit of individuals through the FoV, and the time it may take to triage individuals, the widespread adoption of static camera systems to continuously monitor vital signs has had suboptimal uptake. Robotic systems enable autonomous and continuous monitoring, but these require expensive cameras, computers, and robotic platforms, limiting widespread deployment. In response, we propose a cost-effective and scalable robotic solution consisting of a suite of commercial, off-the-shelf wireless cameras for capturing photoplethysmography (PPG) on ambulatory subjects linked to a single computer that supervises the cameras to compute the vital signs of subjects. Throughout a set of careful investigations of each individual step of the wireless machine vision camera and computer, bottlenecks constraining wireless live-streaming of high-quality PPG information are identified and those are addressed by a hybrid centralized/decentralized wireless machine vision protocol. Our results demonstrate that the proposed cost-effective wireless camera achieves equivalent remote-PPG accuracy to its costly, USB3 counterparts (mean error: 5.0 BMP vs. 4.7 BPM) by means of the hybrid camera protocol which boosts the overall frame rate to 17 FPS. In contrast, using the standard method that captures the PPG with the same spatial resolution can only achieve 1 FPS. In addition, this work also elucidates how varying the distance, image pixel density, frame rate, image compression, image downsampling, and color depth affect the rPPG performance. For each of the effects, we also discuss potential solutions for the cost-effective setup.

As offshore industries and the Internet of underwater things (IoUT) continue to expand, air-to-water communication is needed to connect various underwater sensors and things to the outside world. Visible light communication (VLC) has the potential to provide higher data rates and longer transmission distances than radio frequency and other air-to-water communication techniques. This paper presents an air-to-water VLC system where a light emitting diode (LED) transmitter carried by a hovering unmanned aerial vehicle (UAV) transmits data to an underwater receiver. We minimize the UAV energy consumption that includes both LED transmission and UAV propulsion energy consumption by optimizing the LED transmit power while guaranteeing that the average received signal-tonoise ratio (SNR) at the underwater receiver is above the required SNR. Simulation results show that our optimized LED transmit power reduces the total UAV energy consumption by up to 46% as compared to the benchmark schemes without increasing the transmission time for a given data volume.

This paper presents three new power-efficient designs of ternarySRAM using carbon-nanotube-field-effect transistors (CNTFETs). Two of the proposed ternary SRAM designs are cycle-operator based and the third design is buffer-based.

Awarded “Best Student Paper” and published in Proceedings of The 16th International Conference on Artificial General Intelligence, Stockholm, 2023.To make accurate inferences in an interactive setting, an agent must not confuse passive observation of events with having intervened to cause them. The do operator formalises interventions so that we may reason about their effect. Yet there exist pareto optimal mathematical formalisms of general intelligence in an interactive setting which, presupposing no explicit representation of intervention, make maximally accurate inferences. We examine one such formalism. We show that in the absence of a do operator, an intervention can be represented by a variable. We then argue that variables are abstractions, and that need to explicitly represent interventions in advance arises only because we presuppose these sorts of abstractions. The aforementioned formalism avoids this and so, initial conditions permitting, representations of relevant causal interventions will emerge through induction. These emergent abstractions function as representations of one’s self and of any other object, inasmuch as the interventions of those objects impact the satisfaction of goals. We argue that this explains how one might reason about one’s own identity and intent, those of others, of one’s own as perceived by others and so on. In a narrow sense this describes what it is to be aware, and is a mechanistic explanation of aspects of consciousness.

Modular Multilevel Cascade Converter (MMCC) have achieved a great importance in high voltage converter applications. As they require a high amount of capacitors whose voltages must be known for the control of the MMCC, many voltage sensors are needed. This paper presents a nonlinear Lyapunov-based observer for cell voltage estimation. The proposed observer does solely rely on the knowledge of the phase currents i.e. no voltage sensors are required and the convergence of the estimator is shown with a stability proof by Lyapunov. The performance of the observer is validated simulatively using MATLAB/Simulink.

Knowledge Tracing is a well-known problem often tackled through purely data-driven approaches. In recent years, many advances have been made in research with various machine learning and deep learning techniques. Despite their satisfactory performance, they have some pitfalls, e.g. modeling one skill at a time, ignoring the relationships between different skills, or inconsistency for the predictions, i.e. sudden spikes and falls across time steps. For this reason, hybrid machine-learning techniques have also been explored. With this systematic literature review, we aim to illustrate this field’s state of the art. Specifically, we want to identify the potential and the frontiers in integrating prior knowledge sources in the traditional machine learning pipeline as a supplement to the normally considered data. We applied a qualitative analysis to distill a taxonomy with three dimensions: knowledge source, knowledge representation, and knowledge integration. Exploiting this taxonomy, we also conducted a quantitative analysis to detect the most common approaches.

Seismic AVO has a significant potential for fluid identification in time-lapse monitoring of the cyclic recovery of geothermal reservoirs. With this goal, we develop an AVO method based on the reflection and transmission (R/T) of elastic waves at an interface between two fluid-saturated thermo-poroelastic media. The method is applied to the Olkaria geothermal reservoirs in Kenya. This system is characteristic of a natural cyclic recovery, where cyclic meteoric water undergoes complex phase transition and thermo-hydro-mechanical coupling process. Conceptual models are built based on petrophysical and thermophysical properties of trachyte thermal reservoirs in the eastern field, with an attempt to model the shallow steam and deep boiling water zones. A plane-wave analysis illustrates the effects of thermal conductivity, specific heat, and porosity on velocity dispersion and attenuation of the fast-P, Biot slow-P, and thermal slow-P waves. AVO modeling by P-wave incidence is conducted to investigate the effects of temperature, porosity, specific heat, and fluid type on the R/T coefficients. For trachyte reservoirs with a temperature less than 400°C, limited changes in the thermophysical properties (e.g., thermal conductivity and specific heat) have negligible effects on wave propagation, whereas significant effects are due to temperature, porosity, and fluid type. Particularly, comparisons of cyclic recovery using water, supercritical CO2, and gas (dry case) as the heat transfer fluid, demonstrate that the crossplot of fluid factors and intercept gradient (PG) can be used as a precursor to hydrofracturing-induced permeability, fluid leakage or short circuits.

The design and development of sub-Terahertz (THz) communication systems entail the need for new channel models that can precisely predict channel characteristics at such frequencies (>100 GHz) in outdoor and dynamic environments. This work proposes a novel hybrid-stochastic ultra-wideband channel model for sub-THz bands, developed within the framework of a class of spatio-temporal stochastic processes called ambit-process. The proposed model is capable of supporting bandwidths of upto 10 GHz. The spatio-temporal evolution of the ambit framework allows for a reasonably accurate and tractable characterization of the fading statistics and multipath propagation of a cellular channel with good spatial consistency. We leverage a recently proposed low-complexity ambit-process based channel simulation algorithm with necessary modifications to study key features of the outdoor sub-THz channel like associated diffused scattering, molecular absorption, spatio-temporal correlations, and consistency between the time-evolving delay and Doppler of the multipaths. Simulation results on path loss, delay spread, and channel correlation indicate that the ambit model accurately captures the typicalities of an urban microcellular sub-THz channel and agrees well with the measurement results reported in the literature.

Noise can be modeled as a sequence of random variables defined on a probability space that may add to a given dynamical system T that is a map on a phase space. In the nontrivial case of dynamical noise {εn}n, where εn ∈ N(0, σ2) and the system output is xn = T(xn−1, xn−2, …, x0)+εn, without any specific knowledge/assumption on T, the quantitative estimation of noise power σ2 is a challenge. Here, we introduce a formal method, based on nonlinear entropy profile, to estimate the dynamical noise power σ2 without knowledge on the specific T function. Time-series generated from Logistic maps and Pomeau-Manneville systems under different conditions are used to test the correctness of the method. Results demonstrate that the proposed estimation algorithm properly discerns different noise levels with no a priori information. Numerical simluations, including sereis generation, are performed with MATLAB.

In the bidding process in the power generation market, enterprises often exhibit speculative behaviors due to the loopholes in bidding rules. To explore the evolution of power generation enterprises from a benign bidding state to a speculative bidding state, an evolutionary game model is established. Based on the replicator dynamics equation, an evolutionary game model of a power generation market containing multiple power generation enterprise groups with different bidding strategies is constructed by introducing the relationship factors of the strategies. Then, to analyze the internal mechanism of the outbreak of speculative behavior, the replicator dynamics equation is transformed into a classical sharp point mutation model. On this basis, Gaussian white noise is introduced to construct the stochastic catastrophe model. The influence of the random disturbance term on the stability of the bidding behavior of enterprises is discussed from the perspective of the trajectory difference between the analytical solution and the numerical solution of the stochastic differential equation. The results show that the final decision state of the group strategy and the convergence time of group evolution to this state depend on the value of the strategy relationship factor. The emergence of random disturbances will lead to heterogeneous decision-making actions within the group. That is, the evolution of the group will fluctuate near the strategic equilibrium point. The difference between the analytical solution and the numerical solution of the stochastic catastrophe model can increase the possibility of a sudden change in the bidding strategy state and thereby decrease the stability of the bidding behavior.