Estimating air-to-ground (A2G) channel for a cellular-connected unmanned aerial vehicle (UAV) requires frequent pilot transmission due to its high mobility. To reduce the pilot overhead, researchers have attempted to predict future channels based on the historical ones by leveraging learning techniques. However, most existing works are limited to sequentially forecasting subsequent channels, suffering from the error accumulation problem that consequently hampers the prediction accuracy. To address this issue, this paper proposes a novel learning-based estimate-then-predict scheme for A2G channel tracking. In this scheme, the UAV transmits limited pilots, and the base station (BS) first performs channel estimation by exploiting the received pilots and then predicts a series of subsequent channels concurrently. Specifically, in the estimation phase, we propose a least-squares feedforward neural network (LS-FNN) to fuse the benefits of LS in high signal-to-noise ratio (SNR) regime and FNN in low SNR regime. In the prediction phase, a multi-time-interval long-short-term-memory (MTI-LSTM) network is proposed for concurrent channel prediction. A distinctive difference from prior works is that layer normalization is employed to greatly increase the prediction accuracy at no cost of additional neurons. Simulation results corroborate the superior performance of our proposed scheme over the state-of-the-art benchmarks.

Protein solubility plays a crucial role in various biotechnological, industrial, and biomedical applications. With the reduction in sequencing and gene synthesis costs, the adoption of high-throughput experimental screening coupled with tailored bioinformatic prediction has witnessed a rapidly growing trend for the development of novel functional enzymes of interest (EOI). High protein solubility rates are essential in this process and accurate prediction of solubility is a challenging task. As deep learning technology continues to evolve, attention-based protein language models (PLMs) can extract intrinsic information from protein sequences to a greater extent. Leveraging these models along with the increasing availability of protein solubility data inferred from structural database like the Protein Data Bank holds great potential to enhance the prediction of protein solubility. In this study, we curated an Updated Escherichia coli protein Solubility DataSet (UESolDS) and employed a combination of multiple PLMs and classification layers to predict protein solubility. The resulting best-performing model, named Protein Language Model-based protein Solubility prediction model (PLM_Sol), demonstrated significant improvements over previous reported models, achieving a notable 6.4% increase in accuracy, 9.0% increase in F1_score, and 11.1% increase in Matthews correlation coefficient score on the independent test set. Moreover, additional evaluation utilizing our in-house synthesized protein resource as test data, encompassing diverse types of enzymes, also showcased the good performance of PLM_Sol. Overall, PLM_Sol exhibited consistent and promising performance across both independent test set and experimental set, thereby making it well suited for facilitating large-scale EOI studies. PLM_Sol is available as a standalone program and as an easy-to-use model at https://zenodo.org/doi/10.5281/zenodo.10675340.

The evaluation and identification of high-quality patents are urgently needed for the technological research and development and the transformation of achievements. Traditional researchers and analysts mainly focus on developing various patent quality indicators.However, there is a lack of relative research on how to apply these indicators to comprehensively evaluate patent quality. Therefore, this paper uses the rough set theory(RST) and multidimensional cloud model(MCM) to construct a comprehensive evaluation and grading system for patent quality (RST-MCM), which is used to comprehensively evaluate the quality of patents. First, by systematically summarizing the relevant literature on patent quality evaluation, we identify the influencing factors of patent quality in multiple dimensions and at multiple stages. Second, the patent quality evaluation index system is constructed by using RST to reduce redundant patent quality influencing factors and determine the evaluation index weights. Finally, the evaluation and grading of patent quality is completed with MCM. To validate the effectiveness of the research, RST-MCM is applied to the quality evaluation of patents in the construction engineering industry. The research results show that the accuracy rate of RST-MCM is 90.3%. The research will provide effective decision-making support for the formulation of technical strategies such as improving independent innovation capability and patent layout.

Most advances in neural radiance fields (NeRF) assume sufficient input views from pinhole cameras. This paper proposes rpcPRF, a Multiplane Images (MPI) based Planar neural Radiance Field for Rational Polynomial Camera (RPC), exhibits robust performance even with single or sparse inputs. Unlike coordinate-based NeRFs that require sufficient views of one scene, our model can be applied to new scenes and has shown positive results for single or sparse test images. This allows for generalization across various scenes. To achieve generalization across scenes, we use reprojection supervision to ensure that the predicted MPI (multiplane image) accurately captures the geometry between the 3D coordinates and the images. Additionally, we have obviated the requisite for dense depth supervision in multiview-stereo-based methods by introducing rendering techniques of radiance fields. rpcPRF combines the superiority of implicit representations and the advantages of the RPC model, to capture the continuous altitude space while learning the 3D structure. On the DFC2019 dataset with sparse input of the same scene, rpcPRF achieves the best results. On the TLC dataset and the SatMVS3D dataset with changing scenes in every batch, rpcPRF outperforms state-of-the-art NeRF-based methods by a significant margin in terms of image fidelity, reconstruction accuracy, and efficiency, for both single-view and multiview task.

Existing NeRF models for satellite imagery have limitations in processing large images and require solar input, leading to slow speeds. As a response, we introduce SatensoRF, which speeds up the entire process significantly while using fewer parameters for large satellite imagery. We have noticed that the common assumption of Lambertian surfaces in satellite neural radiance fields is not sufficient for vegetative and aquatic elements. In contrast to the traditional hierarchical MLP-based scene representation, we have chosen a multiscale tensor decomposition approach for color, volume density, and auxiliary variables to model the light field with specular color. Additionally, to rectify inconsistencies in multi-date imagery, we incorporate total variation denoising to restore the density tensor field, thus mitigating the negative impact of transient objects. To validate our approach, we conducted assessments of SatensoRF using subsets from the spacenet multi-view dataset, which includes both multi-date and single-date multi-view RGB images. Our results demonstrate that SatensoRF surpasses the state-of-the-art Sat-NeRF series regarding novel view synthesis performance. Significantly, SatensoRF requires fewer parameters for training, resulting in faster training and inference speeds and reduced computational demands.

Patient similarity assessment (PSA) is pivotal to evidence-based and personalized medicine, enabled by analyzing the increasingly available electronic health records (EHRs). However, machine learning approaches for PSA have to deal with inherent data deficiencies of EHRs, namely missing values, noise, and small sample sizes. In this work, an end-to-end discriminative learning framework, called SparGE, is proposed to address these data challenges of EHR for PSA. SparGE measures similarity by jointly sparse coding and graph embedding. First, we use low-rank constrained sparse coding to identify and calculate weight for similar patients, while denoising against missing values. Then, graph embedding on sparse representations is adopted to measure the similarity between patient pairs via preserving local relationships defined by distances. Finally, a global cost function is constructed to optimize related parameters. Experimental results on two private and public real-world healthcare datasets, namely SingHEART and MIMIC-III, show that the proposed SparGE significantly outperforms other machine learning patient similarity methods.

MOTIVATION: Mass cytometry (Cytometry by Time-Of-Flight, CyTOF) is a single-cell technology that is able to quantify multiplex biomarker expressions and is commonly used in basic life science and translational research. However, the widely used Gadolinium (Gd)-based contrast agents (GBCAs) in magnetic resonance imaging (MRI) scanning in clinical practice can lead to signal contamination on the Gd channels in the CyTOF analysis. This Gd contamination greatly affects the characterization of the real signal from Gd-isotope-conjugated antibodies, severely impairing the CyTOF data quality and ruining downstream single-cell data interpretation. RESULTS: We first in-depth characterized the signals of Gd isotopes from a control sample that was not stained with Gd-labeled antibodies but was contaminated by Gd isotopes from GBCAs, and revealed the collinear intensity relationship across Gd contamination signals. We also found that the intensity ratios of detected Gd contamination signals to the reference Gd signal were highly correlated with the natural abundance ratios of corresponding Gd isotopes. We then developed a computational method named by GdClean to remove the Gd contamination signal at the single-cell level in the CyTOF data. We further demonstrated that the GdClean effectively cleaned up the Gd contamination signal while preserving the real Gd-labeled antibodies signal in Gd channels. All of these shed lights on the promising applications of the GdClean method in preprocessing CyTOF datasets for revealing the true single-cell information. AVAILABILITY AND IMPLEMENTATION: The R package GdClean is available on GitHub at https://github.com/JunweiLiu0208/GdClean. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Microcredit is a new financial instrument serving the segment of population that typically lack collateral and are highly likely to be rejected by traditional financial institutions, by lending very small loans. For platforms that participate in such consumer finance activities, the key challenge lies in risk management and the popular credit scoring method predicting whether a borrower would default or not takes an important role in this field. However, the fact is that we are often facing mass of raw data and the traditional credit scoring is heavily depending on feature engineering involving domain expert knowledge, intuition and trial and error, which is often time consuming. It is very challenging to derive effective features from raw data as the searching space can be very large with noninformative features. In this paper, we propose a new performance-driven framework automated generating discriminating features from raw data via reinforcement learning to help improve the default prediction of the downstream classifier which may be a logistic regression or boosting tree. Specially, we first define a formal paradigm for the automated feature derivation framework which unifies the feature structure, its interpretation and the calculation logic together. For the particularity of the financial industry, the interpretation of the feature is often of high interest. Then we reformulate the feature generation problem as reinforcement learning by constructing a transformation link and regarding it as a sequential decision process. In addition, we carry out an effective practice on default prediction in consumer finance. Finally, experimental results on the data of user behavior log from 360 Financial show the significant improvement of the proposed method over our years of domain expert knowledge and the Genetic Programming. Moreover, this FDRL framework can be easily adapted to other applications due to its versatility.

Sixteen conventional heart beat variability (HRV) parameters and eight vital signs have shown promise in the prediction of cardiac arrest within 72 hours. Besides these 24 parameters, we proposed adding two new features for cardiac arrest prediction, which are approximate entropy (ApEn) and sample entropy (SpEn). ApEn and SpEn are nonlinear HRV parameters capable of characterizing heart conditions. These two entropies were derived from electrocardiography recordings and combined with the existing 24 features to form feature combinations. The experiments were conducted by using linear kernel Support Vector Machine classification technique to investigate the effects of using ApEn, SpEn together with 24 parameters on cardiac arrest prediction. The dimensionality reduction approach, Principal Component Analysis, was applied to suppress the dimensionality. Results reveal that the prediction performance of adding ApEn and SpEn to the 24 parameters is improved significantly compared to using the 24 parameters only. Dimension reduction has additional positive effects on improving the prediction results.

Traditional risk score prediction is based on vital signs and clinical assessment. In this paper, we present an intelligent scoring system for the prediction of cardiac arrest within 72 h. The patient population is represented by a set of feature vectors, from which risk scores are derived based on geometric distance calculation and support vector machine. Each feature vector is a combination of heart rate variability (HRV) parameters and vital signs. Performance evaluation is conducted on the leave-one-out cross-validation framework, and receiver operating characteristic, sensitivity, specificity, positive predictive value, and negative predictive value are reported. Experimental results reveal that the proposed scoring system not only achieves satisfactory performance on determining the risk of cardiac arrest within 72 h but also has the ability to generate continuous risk scores rather than a simple binary decision by a traditional classifier. Furthermore, the proposed scoring system works well for both balanced and imbalanced datasets, and the combination of HRV parameters and vital signs shows superiority in prediction to using HRV parameters only or vital signs only.

This paper presents an alternative approach for heart rate measurement. Instead of using an ECG sensor, the proposed design uses sound signals received from a microphone which does not require skin-contact. Specifically, the design uses an air conductive microphone and an efficient algorithm to estimate the heart beat parameters of the wearer. The estimates are obtained for different activities undertaken by the wearer. The activities studied include sitting, jumping, reading, laughing, singing, and coughing. Data are collected from researchers and students working in the laboratory, and in the presence of lung sound, and other environmental sounds and noise. Preliminary results show that, the method can be an effective alternative means of monitoring cardiac (heart) sounds in a natural environment.

The minimum rectilinear Steiner tree (MRST) problem arises in global routing and wiring estimation, as well as in many other areas. The MRST problem is known to be NP-hard, and the best performing MRST heuristic to date is the Iterated 1-Steiner (I1S) method recently proposed by Kahng and Robins (see ibid., vol. 11, p. 893-902, 1992). In this paper, we develop a straightforward, efficient implementation of I1S, achieving a speedup factor of three orders of magnitude over previous implementations. We also give a parallel implementation that achieves near-linear speedup on multiple processors. Several performance-improving enhancements enable us to obtain Steiner trees with average cost within 0.25% of optimal, and our methods produce optimal solutions in up to 90% of the cases for typical nets. We generalize I1S and its variants to three dimensions, as well as to the case where all the pins lie on k parallel planes, which arises in, e.g., multilayer routing. Motivated by the goal of reducing the running times of our algorithms, we prove that any pointset in the Manhattan plane has a minimum spanning tree (MST) with maximum degree 4, and that in three-dimensional Manhattan space every pointset has an MST with maximum degree of 14 (the best previous upper bounds on the maximum MST degree in two and three dimensions are 6 and 26, respectively); these results are of independent theoretical interest and also settle an open problem in complexity theory.>

The minimum rectilinear Steiner tree (MRST) problem is known to be NP-hard, and the best performing MRST heuristic to date is the Iterated 1-Steiner (I1S) method recently proposed by A.B. Kahng and G. Robins (1992). The authors develop a straightforward, efficient implementation of I1S, achieving speedup factors of over 200 compared to previous implementations. They also propose a parallel implementation of I1S that achieves high parallel speedup on K processors. Extensive empirical testing confirms the viability of the approach, which allows the benchmarking of I1S on nets containing several hundred pins.>