Search

In this paper, we present a microwave photonic (MWP) sensor whose operational range and sensing accuracy are enhanced through the utilization of a recurrent neural network (RNN). The MWP sensor utilizes an optical microresonator as the sensor probe and converts the optical resonance responses near the optical carrier frequency into variations in RF transmission with high interrogation resolution. To overcome bandwidth limitations and achieve wide-range operation, a tunable laser is employed to perform the high-resolution interrogation across multiple optical carrier frequencies during each measurement cycle. Subsequently, a RNN, leveraging long-range dependencies and shared parameters, is integrated to process the concatenated interrogation outputs after dimensionality reduction, compensating for output wavelength discrepancies of the tunable laser and enabling accurate wide-range sensing. The proposed approach is experimentally validated using a microring resonator to measure fructose solution concentrations while contending with laser frequency deviation and thermal interference. The operational range of the system is extended three times to 114 GHz, facilitating the measurement of solution concentrations ranging from 49.91% to 30.43% under a temperature variation of 0.61 °C and a laser frequency deviation of ±2 GHz. The established RNN model demonstrates a root-mean-square error of 0.11%, showcasing 1.60-fold, 2.77-fold, 1.10-fold, and 3.45-fold improvements in accuracy over models based on convolutional neural networks, multilayer perceptrons, sparse vision transformer, and linear fitting, respectively.

Microwave photonic (MWP) sensors, facilitating high speed and high resolution through the conversion of optical responses from the optical to the radio frequency domain, are becoming indispensable in an era where advanced sensing capabilities are paramount. The combination of machine learning (ML) with microwave photonics has provided advanced solutions that were previously unattainable. This paper begins by elucidating the operational principles of MWP sensing, and then proceeds to present the latest developments in the merging of ML and deep learning (DL) with integrated MWP sensors, where the development of photonic integration enables the realisation of on-chip sensors with significant improvements in both performance and miniaturization. ML/DL assisted MWP sensors with enhanced sensing capabilities, including athermal operation, resistance to noise and interference, multiple parameters detection, and extended sensing range, while exhibiting compactness, cost-effectiveness, and scalability, are presented. Prospective opportunities that could further propel the field of MWP sensing are also discussed.

In this article, a new deep learning enhanced microwave photonic (MWP) sensor based on optical sideband processing with linear frequency modulated (LFM) pulse to enable fast and accurate sensing in the presence of noise and interference, is proposed and demonstrated. The LFM pulse produces a fast sweep of the optical sideband across the resonance, causing an envelope dip in the transmitted LFM pulse which enables fast interrogation. By optimizing the optical modulator bias point, the depth of this envelope dip can be increased, thus enhancing the interrogation resolution and facilitating higher accuracy. To enhance the noise resistance for practical uses and avoid adding unnecessary system complexity, we adopt a deep neural network (DNN) to process the raw interrogation output of this time-domain MWP sensor. The DNN can compress input data into a lower-dimensional representation and be trained to accurately estimate the target measurand despite noise and interference. Beside presenting a detailed theoretical model, as a proof-of-concept, we demonstrate the proposed scheme with a convolutional neural network (CNN) in measuring the glycerol solution concentration amidst thermal interference and system noise. The interrogation speed reached 2.5 MHz. Compared to the conventional MWP interrogation without bias control, using optimized bias voltages improves the accuracy of the estimation model based on the linear fitting of the manually extracted envelope dip by 2-fold. The CNN based prediction model achieves a root-mean-square error of 0.05%, which demonstrates an overall 4-fold higher accuracy than that of the reference model using traditional interrogation and filtering methods.

The combination of optical microresonators and the emerging microwave photonic (MWP) sensing has recently drawn great attention, whereas its multi-parameter sensing capability mainly relies on adopting multiple resonance modes. By incorporating deep learning (DL) into MWP sensing, we propose a new sensing paradigm, which has the simplified design, reduced fabrication requirement, and the capability of sensing more than one parameter. The MWP interrogation transforms the spectral response of a single optical resonance (SOR) that can be at arbitrary coupling conditions into the variations of the zero-transmission profile of microwave signals, providing improved interrogation resolution regardless of the resonance parameters. A DL unit is used to exploit the raw interrogation output to simultaneously estimate the target measurands. As the proof-of-concept demonstration, simultaneous temperature and humidity sensing using a SOR is conducted, where the convolutional neural tangent kernel (CNTK) is used as the DL model to reduce the demand for experimental data. The established CNTK-DL model consistently outperforms the support vector regression model that relies on handcrafted features and demonstrates an over 2-fold higher estimation accuracy with the laser drift interference and a lower mean absolute error in the presence of strong noise, showing the power of DL for boosting MWP sensing.

We propose a machine learning (ML) assisted athermal microwave photonic (MWP) sensing scheme with high resolution based on a single microring resonance. The immunity of temperature interference of the high-resolution sensing is achieved by employing MWP sideband processing based interrogation, and supervised machine learning based on support vector regression (SVR) and neural tangent kernel (NTK) that are effective on small datasets. The MWP sideband processing transforms the variation of the target measurand into the shift of an ultra-deep notch in the radio frequency (RF) spectrum relieving the fabrication requirements on the microresonators, while ML accurately predicts the measurand by using the modulator bias voltage or RF passband transmission together with the RF notch position. The proposed sensor is demonstrated for relative humidity (RH) measurements using a silicon-on-insulator microring coated with polymethyl methacrylate. About 50 dB high RF rejection ratio is achieved over the sensing process, indicating a high sensing resolution. Despite the small experimental datasets, the established SVR and NTK models consistently exhibit lower mean absolute errors (MAEs) than the linear regression model in the RH prediction in the presence of temperature drifts. The NTK models achieve the lowest MAEs of 1.01% RH and 1.03% RH when the RF passband transmission and modulator bias are selected as the model input, respectively. The equivalent performances of the RF passband transmission and modulator bias voltage further demonstrate the feasibility of athermal sensing based solely on the MWP interrogation results of a single microring resonance, which simplifies the design and reduces the complexity.

Sensors stand as pivotal cornerstones of technology, driving progress across a spectrum of industries through their ability to precisely capture and interpret an extensive array of physical phenomena. Among these advancements, microwave photonic (MWP) sensing has emerged as a new sensing technique, elevating sensing speed and resolution for practical applications. Integrated MWP sensors exhibit unparalleled capabilities in ultra‐sensitive, label‐free nanoscale detection, offering the potential to synergize with advanced integration techniques for a compact footprint and versatile designs. This paper reviews and summarizes the development and recent advances in integrated MWP sensing, focusing on the schemes based on microresonators. The diverse array of existing schemes is systematically categorized, elucidating their operational principles and performance demonstration. Furthermore, the assistance of machine learning and deep learning in integrated MWP sensors is explored, highlighting the potential of intelligent sensing paradigms. Finally, current challenges and opportunities aimed at further advancing MWP sensors are discussed. This manuscript reviews the recent advancements in integrated microwave photonic sensors (IMWPS). It categorizes the diverse array of existing schemes, introduces their operational principles, demonstrates their effectiveness in sensing performance, and explores the assistance of machine learning and deep learning to enhance sensing capabilities. Additionally, opportunities that can further propel the field of IMWPS are also discussed.

This paper deals mainly with some aspects of the adjointable operators on Hilbert C∗-modules. A new tool called the generalized polar decomposition for each adjointable operator is introduced and clarified. As an application, the general theory of the weakly complementable operators is set up in the framework of Hilbert C∗-modules. It is proved that there exists an operator equation which has a unique solution, whereas this unique solution fails to be the reduced solution. Some investigations are also carried out in the Hilbert space case. It is proved that there exist a closed subspace Mof certain Hilbert space Kand an operator T∈B(K)such that Tis (M, M)-weakly complementable, whereas Tfails to be (M, M)-complementable. The solvability of the equation A:B=X∗AX+(I-X)∗B(I-X)(X∈B(H))is also dealt with in the Hilbert space case, where A,B∈B(H)are two general positive operators, and A: Bdenotes their parallel sum. Among other things, it is shown that there exist certain positive operators Aand Bon the Hilbert space ℓ2(N)⊕ℓ2(N)such that the above equation has no solution.

As artificial intelligence (AI) becomes more prominent in children's lives, an increasing number of researchers and practitioners have underscored the importance of integrating AI as learning content in K-12. Despite the recent efforts in developing AI curricula and guiding frameworks in AI education, the educational opportunities often do not provide equally engaging and inclusive learning experiences for all learners. To promote equality and equity in society and increase competitiveness in the AI workforce, it is essential to broaden participation in AI education. However, a framework that guides teachers and learning designers in designing inclusive learning opportunities tailored for AI education is lacking. Universal Design for Learning (UDL) provides guidelines for making learning more inclusive across disciplines. Based on the principles of UDL, this paper proposes a framework to guide the design of inclusive AI learning. We conducted a systematic literature review to identify AI learning design-related frameworks and synthesized them into our proposed framework, which includes the core component of AI learning content (i.e., five big ideas), anchored by the three UDL principles (the “why,” “what,” and “how” of learning), and six praxes with pedagogical examples of AI instruction. Alongside this, we present an illustrative example of the application of our proposed framework in the context of a middle school AI summer camp. We hope this paper will guide researchers and practitioners in designing more inclusive AI learning experiences.

This study examined undergraduate STEM students’ experiences using an online introductory computer programming learning environment equipped with an automated hint generation system. Following a convergent parallel mixed methods design, this study utilized both quantitative and qualitative data from student experiential data. Analysis by level of prior knowledge demonstrated that elements of the learning environment did not cater to their learning needs and cognitive architecture. Cognitive Load Theory was used to contextualize system elements against both higher and lower prior experience learners, ultimately pointing to a need to design better scaffolds and hints to the needs of novice CS learners.

Conversational AIs such as Alexa and ChatGPT are increasingly ubiquitous in young people’s lives, but these young users are often not afforded the opportunity to learn about the inner workings of these technologies. One of the most powerful ways to foster this learning is to empower youth to create AI that is personally and socially meaningful to them. We have built a novel development environment, AMBY–“AI Made By You”–for youth to create conversational agents. AMBY was iteratively designed with and for youth aged 12–13 through contextual inquiry and usability studies. AMBY is designed to foster AI learning with features that enable users to generate training datasets and visualize conversational flow. We report on results from a two-week summer camp deployment, and contribute design implications for conversational AI authoring tools that empower AI learning for youth.

Suppose that pand qare projections in a unital C∗-algebra Asuch that ‖p(1-q)‖<1. It is shown that there exists a unitary uin Awhich is homotopic to the unit of A, and satisfies pup=pu∗p, u(pqp)u∗=qpqand ‖1-u‖≤2‖(qp)†‖1+‖(qp)†‖·‖p(1-q)‖, where (qp)†denotes the Moore–Penrose inverse of qp. Under the same restriction of ‖p(1-q)‖<1, it is proved that ‖p-q‖<1if and only if there exists a unitary uin Asuch that pupis normal and q=upu∗. An example is constructed to show that there exist certain Hilbert space Hand projections pand qon Hsuch that ‖p-q‖=1and q=upu∗for some unitary operator uon H.

Clinical or epidemiologic screening of single-nucleotide polymorphism markers requires large-scale multiplexed genotyping. Available genotyping tools require DNA extraction and multiplex PCR, which may limit throughput and suffer amplification bias. Herein, a novel genotyping approach has been developed, multiplex extension and ligation-based probe amplification (MELPA), which eliminates DNA extraction and achieves uniform PCR amplification. MELPA lyses blood or dried blood spot and directly captures specific target DNA to 96-well plates using tailed probes. Subsequent enzymatic extension and ligation form target single-nucleotide polymorphism–spanning single-stranded templates, which are PCR-amplified using universal primers. Multiplexed genotyping by single-base primer extension is analyzed by mass spectrometry, with a call rate >97%. MELPA was compared with a commercial assay (iPLEX) for detecting 24 G6PDvariants known to be at risk for primaquine-induced hemolysis. MELPA provided results that were more reliable than iPLEX, with higher throughput and lower cost. Genotyping archival blood from 106 malaria patients taking primaquine found 10 G6PD-deficient variants, including 1 patient with a hemizygous Mahidol mutation who had hemolysis. Preemptive G6PDgenotyping of 438 dried blood spots from a malaria-endemic area identified three variants. MELPA also enabled pooled genotyping without diluting rare alleles, in which undesired common-allele background increased by sample pooling can be repressed by adding specific common allele blockers. Thus, MELPA represents a high-throughput, cost-effective approach to targeted genotyping at the population level.

In this paper, we will investigate the prescribed-time tracking with prescribed performance for a class of strict-feedback nonlinear systems. A control strategy will be proposed to achieve the tracking error constrain by prescribed performance bound (PPB) and converge to zero in prescribed-time. The PPB constrains not only the convergence rate, but also the maximum overshoot of the tracking error. The tracking error will converge to zero in prescribed-time. Such a prescribed-time is irrespective of both initial conditions and design parameters. First, by incorporating the PPB into the original system via a transformation, a equivalently transformed system is obtained, which is unconstrained and convenient for control design. Then, with the help of backstepping, a controller is recursively designed to guarantee that the tracking error satisfies the PPB all the time and converges to zero in prescribed-time. The key to designing the controller is that a fractional term is introduced when designing the virtual stabilizing function during each step of backstepping, which in turn decreases the Lyapunov function to origin within prescribed-time. Finally, two examples are presented to demonstrate the validity of our proposed control strategy. [ABSTRACT FROM AUTHOR]