Learning Mechanism Research Articles

Ualign: pushing the limit of template-free retrosynthesis prediction with unsupervised SMILES alignment

MotivationRetrosynthesis planning poses a formidable challenge in the organic chemical industry, particularly in pharmaceuticals. Single-step retrosynthesis prediction, a crucial step in the planning process, has witnessed a surge in interest in recent years due to advancements in AI for science. Various deep learning-based methods have been proposed for this task in recent years, incorporating diverse levels of additional chemical knowledge dependency.ResultsThis paper introduces UAlign, a template-free graph-to-sequence pipeline for retrosynthesis prediction. By combining graph neural networks and Transformers, our method can more effectively leverage the inherent graph structure of molecules. Based on the fact that the majority of molecule structures remain unchanged during a chemical reaction, we propose a simple yet effective SMILES alignment technique to facilitate the reuse of unchanged structures for reactant generation. Extensive experiments show that our method substantially outperforms state-of-the-art template-free and semi-template-based approaches. Importantly, our template-free method achieves effectiveness comparable to, or even surpasses, established powerful template-based methods.Scientific contributionWe present a novel graph-to-sequence template-free retrosynthesis prediction pipeline that overcomes the limitations of Transformer-based methods in molecular representation learning and insufficient utilization of chemical information. We propose an unsupervised learning mechanism for establishing product-atom correspondence with reactant SMILES tokens, achieving even better results than supervised SMILES alignment methods. Extensive experiments demonstrate that UAlign significantly outperforms state-of-the-art template-free methods and rivals or surpasses template-based approaches, with up to 5% (top-5) and 5.4% (top-10) increased accuracy over the strongest baseline.

Eligibility traces in an autonomous soccer robot with obstacle avoidance and navigation policy

As the thematic literature of machine learning suggests, reinforcement learning falls between the two methods of supervised learning and unsupervised learning. In this method, the learning agent receives a reward or punishment from the environment according to its action. Therefore, the learning agent interacts with the environment through trial and error and learns to choose the optimal action to achieve its goal. In the meantime, the eligibility traces are considered as one of the main mechanisms of reinforcement learning in receiving delayed rewards. In the use of conventional reinforcement learning methods, when the learning agent achieves a goal, only the value function of the last state-action pair changes, but all the trace states are affected based on the eligibility traces. In other words, the delayed rewards are distributed throughout the trace. Like the ant pheromone effect, this method can increase learning speed empirically to some extent. A soccer robot on the field encounters moving obstacles such as balls, rival robots, and home robots, and fixed obstacles such as gates and flags, so its environment is in a very dynamic state. Therefore, due to the dynamic environment, it is an important issue for autonomous soccer robots to avoid obstacles and should be considered in real play. The main idea of this study is to determine the appropriate traces for the robot to move towards the ball and ultimately to score with the approach of avoiding obstacles in simulating a real soccer match. The desired results obtained from a game played indicate a high level of online experience and decision-making power in the face of new situations.

Multi-Resolution Learning and Semantic Edge Enhancement for Super-Resolution Semantic Segmentation of Urban Scene Images.

Super-resolution semantic segmentation (SRSS) is a technique that aims to obtain high-resolution semantic segmentation results based on resolution-reduced input images. SRSS can significantly reduce computational cost and enable efficient, high-resolution semantic segmentation on mobile devices with limited resources. Some of the existing methods require modifications of the original semantic segmentation network structure or add additional and complicated processing modules, which limits the flexibility of actual deployment. Furthermore, the lack of detailed information in the low-resolution input image renders existing methods susceptible to misdetection at the semantic edges. To address the above problems, we propose a simple but effective framework called multi-resolution learning and semantic edge enhancement-based super-resolution semantic segmentation (MS-SRSS) which can be applied to any existing encoder-decoder based semantic segmentation network. Specifically, a multi-resolution learning mechanism (MRL) is proposed that enables the feature encoder of the semantic segmentation network to improve its feature extraction ability. Furthermore, we introduce a semantic edge enhancement loss (SEE) to alleviate the false detection at the semantic edges. We conduct extensive experiments on the three challenging benchmarks, Cityscapes, Pascal Context, and Pascal VOC 2012, to verify the effectiveness of our proposed MS-SRSS method. The experimental results show that, compared with the existing methods, our method can obtain the new state-of-the-art semantic segmentation performance.

A Multi-strategy Improved Grasshopper Optimization Algorithm for Solving Global Optimization and Engineering Problems

This paper presents a multi-strategy improved grasshopper optimization algorithm (MSIGOA), which aims to address the shortcomings of the grasshopper optimization algorithm (GOA), including its slow convergence, vulnerability to trapping into local optima, and low accuracy. Firstly, to improve the uniformity of the population distribution in the search space, the MSIGOA uses circle mapping for the population initialization. A nonlinear decreasing coefficient is utilized instead of an original linear decreasing coefficient to improve the local exploitation and global exploration capabilities. Then, the modified golden sine mechanism is added during the position update stage to change the single position update mode of GOA and enhance the local exploitation capability. The greedy strategy is added to greedily select the new and old positions of the individual to retain a better position and increase the speed of convergence. Finally, the quasi-reflection-based learning mechanism is utilized to construct new populations to improve population multiplicity and the capability to escape from the local optima. This paper verifies the efficacy of MSIGOA by comparing it with other advanced algorithms on six engineering design problems, CEC2017 test functions, and 12 classical benchmark functions. The experimental results show that MSIGOA performs better than the original GOA and other compared algorithms and has stronger comprehensive optimization capabilities.

The psychological mechanism of self-regulated learning in a technology-enhanced environment

The psychological mechanism of self-regulated learning in a technology-enhanced environment

Higher-order knowledge-enhanced recommendation with heterogeneous hypergraph multi-attention

Recent advancements in recommender systems have focused on integrating knowledge graphs (KGs) to leverage their auxiliary information. The core idea of KG-enhanced recommenders is to incorporate rich semantic information for more accurate recommendations. However, two main challenges persist: i) Neglecting complex higher-order interactions in the KG-based user-item network, potentially leading to sub-optimal recommendations, and ii) Dealing with the heterogeneous modalities of input sources, such as user-item bipartite graphs and KGs, which may introduce noise and inaccuracies. To address these issues, we present a novel Knowledge-enhanced Heterogeneous Hypergraph Recommender System (KHGRec). KHGRec captures group-wise characteristics of both the interaction network and the KG, modeling complex connections in the KG. Using a collaborative knowledge heterogeneous hypergraph (CKHG), it employs two hypergraph encoders to model group-wise interdependencies and ensure explainability. Additionally, it fuses signals from the input graphs with cross-view self-supervised learning and attention mechanisms. Extensive experiments on four real-world datasets show our model's superiority over various state-of-the-art baselines, with an average 5.18% relative improvement. Additional tests on noise resilience, missing data, and cold-start problems demonstrate the robustness of our KHGRec framework. Our model and evaluation datasets are publicly available at https://github.com/viethungvu1998/KHGRec.

Distortions in Judged Spatial Relations in Large Language Models

We present a benchmark for assessing the capability of large language models (LLMs) to discern intercardinal directions between geographic locations and apply it to three prominent LLMs: GPT-3.5, GPT-4, and Llama-2. This benchmark specifically evaluates whether LLMs exhibit a hierarchical spatial bias similar to humans, where judgments about individual locations’ spatial relationships are influenced by the perceived relationships of the larger groups that contain them. To investigate this, we formulated fourteen questions focusing on well-known U.S. cities. Seven questions were designed to challenge the LLMs with scenarios potentially influenced by the orientation of larger geographical units, such as states or countries, whereas the remaining seven targeted locations were less susceptible to such hierarchical categorization. Among the tested models, GPT-4 exhibited superior performance with 55 percent accuracy, followed by GPT-3.5 at 47 percent and Llama-2 at 45 percent. The models showed significantly reduced accuracy on tasks with suspected hierarchical bias. For example, GPT-4’s accuracy dropped to 33 percent on these tasks, compared to 86 percent on others. The models identified the nearest cardinal direction in most cases, however, reflecting their associative learning mechanism, thereby embodying human-like misconceptions. We discuss avenues for improving the spatial reasoning capabilities of LLMs.

A Classification Method for Incomplete Mixed Data Using Imputation and Feature Selection

Data missing is a ubiquitous problem in real-world systems that adversely affects the performance of machine learning algorithms. Although many useful imputation methods are available to address this issue, they often fail to consider the information provided by both features and labels. As a result, the performance of these methods might be constrained. Furthermore, feature selection as a data quality improvement technique has been widely used and has demonstrated its efficiency. To overcome the limitation of imputation methods, we propose a novel algorithm that combines data imputation and feature selection to tackle classification problems for mixed data. Based on the mean and standard deviation of quantitative features and the selecting probabilities of unique values of categorical features, our algorithm constructs different imputation models for quantitative and categorical features. Particle swarm optimization is used to optimize the parameters of the imputation models and select feature subsets simultaneously. Additionally, we introduce a legacy learning mechanism to enhance the optimization capability of our method. To evaluate the performance of the proposed method, seven algorithms and twelve datasets are used for comparison. The results show that our algorithm outperforms other algorithms in terms of accuracy and F1 score and has reasonable time overhead.

AMIKOMNET: Novel Structure for a Deep Learning Model to Enhance COVID-19 Classification Task Performance

Since early 2020, coronavirus has spread extensively throughout the globe. It was first detected in Wuhan, a province in China. Many researchers have proposed various models to solve problems related to COVID-19 detection. As traditional medical approaches take a lot of time to detect the virus and require specific laboratory tests, the adoption of artificial intelligence (AI), including machine learning, might play an important role in handling the problem. A great deal of research has seen the adoption of AI succeed in the early detection of COVID-19 using X-ray images. Unfortunately, the majority of deep learning adoption for COVID-19 detection has the shortcomings of high error detection and high computation costs. In this study, we employed a hybrid model using an auto-encoder (AE) and a convolutional neural network (CNN) (named AMIKOMNET) with a small number of layers and parameters. We implemented an ensemble learning mechanism in the AMIKOMNET model using Adaboost with the aim of reducing error detection in COVID-19 classification tasks. The experimental results for the binary class show that our model achieved high effectiveness, with 96.90% accuracy, 95.06% recall, 94.67% F1-score, and 96.03% precision. The experimental result for the multiclass achieved 95.13% accuracy, 94.93% recall, 95.75% F1-score, and 96.19% precision. The adoption of Adaboost in AMIKOMNET for the binary class increased the effectiveness of the model to 98.45% accuracy, 96.16% recall, 95.70% F1-score, and 96.87% precision. The adoption of Adaboost in AMIKOMNET in the multiclass classification task also saw an increase in performance, with an accuracy of 96.65%, a recall of 94.93%, an F1-score of 95.76%, and a precision of 96.19%. The implementation of AE to handle image feature extraction combined with a CNN used to handle dimensional image feature reduction achieved outstanding performance when compared to previous work using a deep learning platform. Exploiting Adaboost also increased the effectiveness of the AMIKOMNET model in detecting COVID-19.

Challenges of Using Synthetic Data Generation Methods for Tabular Microdata

The generation of synthetic data holds significant promise for augmenting limited datasets while avoiding privacy issues, facilitating research, and enhancing machine learning models’ robustness. Generative Adversarial Networks (GANs) stand out as promising tools, employing two neural networks—generator and discriminator—to produce synthetic data that mirrors real data distributions. This study evaluates GAN variants (CTGAN, CopulaGAN), a variational autoencoder, and copulas on diverse real datasets of different complexity encompassing numerical and categorical attributes. The results highlight CTGAN’s sensitivity to training parameters and TVAE’s robustness across datasets. Scalability challenges persist, with GANs demanding substantial computational resources. TVAE stands out for its high utility across all datasets, even for high-dimensional data, though it incurs higher privacy risks, which is indicative of the curse of dimensionality. While no single model universally excels, understanding the trade-offs and leveraging model strengths can significantly enhance synthetic data generation (SDG). Future research should focus on adaptive learning mechanisms, scalability enhancements, and standardized evaluation metrics to advance SDG methods effectively. Addressing these challenges will foster broader adoption and application of synthetic data.

Effects of psychedelic, DOI, on nucleus accumbens dopamine signaling to predictable rewards and cues in rats.

Psychedelics produce lasting therapeutic responses in neuropsychiatric diseases suggesting they may disrupt entrenched associations and catalyze learning. Here, we examine psychedelic 5-HT2A/2C agonist, DOI, effects on dopamine signaling in the nucleus accumbens (NAc) core, a region extensively linked to reward learning, motivation, and drug-seeking. We measure phasic dopamine transients following acute DOI administration in rats during well learned Pavlovian tasks in which sequential cues predict rewards. We find that DOI (0.0-1.2 mg/kg, i.p.) increases dopamine signals, photometrically measured using GRABDA optical sensor, to rewards and proximal reward cues, but not to the distal cues that predict these events. We determine that the elevated dopamine produced by DOI to reward cues occurs independently of DOI-induced changes in reward value. The increased dopamine associated with predictable reward cues and rewards supports DOI-induced increases in prediction error signaling. These findings lay a foundation for developing psychedelic strategies aimed at engaging error-driven learning mechanisms to disrupt entrenched associations or produce new associations.

Memory-Augmented 3D Point Cloud Semantic Segmentation Network for Intelligent Mining Shovels.

The semantic segmentation of the 3D operating environment represents the key to intelligent mining shovels' autonomous digging and loading operation. However, the complexity of the operating environment of intelligent mining shovels presents challenges, including the variety of scene targets and the uneven number of samples. This results in low accuracy of 3D semantic segmentation and reduces the autonomous operation accuracy of the intelligent mine shovels. To solve these issues, this paper proposes a 3D point cloud semantic segmentation network based on memory enhancement and lightweight attention mechanisms. This model addresses the challenges of an uneven number of sampled scene targets, insufficient extraction of key features to reduce the semantic segmentation accuracy, and an insufficient lightweight level of the model to reduce deployment capability. Firstly, we investigate the memory enhancement learning mechanism, establishing a memory module for key semantic features of the targets. Furthermore, we address the issue of forgetting non-dominant target point cloud features caused by the unbalanced number of samples and enhance the semantic segmentation accuracy. Subsequently, the channel attention mechanism is studied. An attention module based on the statistical characteristics of the channel is established. The adequacy of the expression of the key features is improved by adjusting the weights of the features. This is done in order to improve the accuracy of semantic segmentation further. Finally, the lightweight mechanism is studied by adopting the deep separable convolution instead of conventional convolution to reduce the number of model parameters. Experiments demonstrate that the proposed method can improve the accuracy of semantic segmentation in the 3D scene and reduce the model's complexity. Semantic segmentation accuracy is improved by 7.15% on average compared with the experimental control methods, which contributes to the improvement of autonomous operation accuracy and safety of intelligent mining shovels.

Research on the Construction of Teacher Professional Learning. Community in the Web 3.0 Era

China’s teacher education reform has always advocated the establishment of a teacher-professional learning community to promote teacher professional development and school progress. In the Web 3.0 era, the increasingly open, secure, and fair network has created conditions for the teacher professional learning community to conduct data confirmation. However, in China, the phenomenon of a “silent spiral” has emerged in the professional learning community of teachers, that is, the inferior opinions of teachers in the professional learning community turn to “silent”, and the superior opinions shout, which to a certain extent hinders the creation, protection, and sharing of teacher metadata, affects the development of data right confirmation in the field of education, and slows down the pace of teacher education practice. This study mainly focuses on the value of building a Chinese teacher professional learning community in the Web 3.0 era, the specific difficulties and reasons, and how to develop a Chinese teacher professional learning community in the future. This study mainly adopts the method of documentary review and draws on the “spiral of silence” theory, which regards the expression of opinions and the spread of “silence” as a spiral process of social communication, and influences and restricts public opinion by creating an “opinion environment”. It is found that under the background of the Web 3.0 era, there is an information cocoon in the professional learning community of teachers, and teachers are affected by the diversity of information. Teachers fear independence and generally produce a herd mentality. It is easy to ignore the disadvantageous climate in the professional learning community of teachers, resulting in the continuous silence of teachers. In addition, teacher groups tend to be too stressed and unable to resist the opinion environment. Therefore, the research draws the following conclusions, the construction of teacher professional learning community should adhere to positive guidance, cultivate the correct learning concept, and build the core value concept; Strengthen teachers’ group consciousness, establish a common vision, and make teachers have a sense of belonging to the community; Coordinate the “online + offline”, build a learning mechanism, and salvage the silent voice of the group in the professional learning community of teachers; Provide supporting conditions around the key development issues of the community to promote the growth of opinion climate.

State-of-the-art review on the use of AI-enhanced computational mechanics in geotechnical engineering

Significant uncertainties can be found in the modelling of geotechnical materials. This can be attributed to the complex behaviour of soils and rocks amidst construction processes. Over the past decades, the field has increasingly embraced the application of artificial intelligence methodologies, thus recognising their suitability in forecasting non-linear relationships intrinsic to materials. This review offers a critical evaluation AI methodologies incorporated in computational mechanics for geotechnical engineering. The analysis categorises four pivotal areas: physical properties, mechanical properties, constitutive models, and other characteristics relevant to geotechnical materials. Among the various methodologies analysed, ANNs stand out as the most commonly used strategy, while other methods such as SVMs, LSTMs, and CNNs also see a significant level of application. The most widely used AI algorithms are Artificial Neural Networks (ANN), Random Forest (RF), and Support Vector Machines (SVM), representing 35%, 19%, and 17% respectively. The most extensive AI application is in the domain of mechanical properties, accounting for 59%, followed by other applications at 16%. The efficacy of AI applications is intrinsically linked to the type of datasets employed, the selected model input. This study also outlines future research directions emphasising the need to integrate physically guided and adaptive learning mechanisms to enhance the reliability and adaptability in addressing multi-scale and multi-physics coupled mechanics problems in geotechnics.

Widened Attention‐Enhanced Atrous Convolutional Network for Efficient Embedded Vision Applications under Resource Constraints

Onboard image analysis enables real‐time autonomous capabilities for unmanned platforms including aerial, ground, and aquatic drones. Performing classification on embedded systems, rather than transmitting data, allows rapid perception and decision‐making critical for time‐sensitive applications such as search and rescue, hazardous environment exploration, and military operations. To fully capitalize on these systems’ potential, specialized deep learning solutions are needed that balance accuracy and computational efficiency for time‐sensitive inference. This article introduces the widened attention‐enhanced atrous convolution‐based efficient network (WACEfNet), a new convolutional neural network designed specifically for real‐time visual classification challenges using resource‐constrained embedded devices. WACEfNet builds on EfficientNet and integrates innovative width‐wise feature processing, atrous convolutions, and attention modules to improve representational power without excessive overhead. Extensive benchmarking confirms state‐of‐the‐art performance from WACEfNet for aerial imaging applications while remaining suitable for embedded deployment. The improvements in accuracy and speed demonstrate the potential of customized deep learning advancements to unlock new capabilities for unmanned aerial vehicles and related embedded systems with tight size, weight, and power constraints. This research offers an optimized framework, combining widened residual learning and attention mechanisms, to meet the unique demands of high‐fidelity real‐time analytics across a variety of embedded perception paradigms.

You Can Make it Too Easy: Simulating Easy-To-Hard Effects in Auditory Perceptual Learning with an Unsupervised Neural Network Model of Plasticity

ABSTRACT The easy-to-hard effect is a phenomenon where training on initially easy perceptual discrimination contrasts before introducing hard contrasts yields greater perceptual learning than training with hard contrasts for the same amount of time. However, easy-to-hard effects can be erased if initial easy trials are made too easy. Here, we show that this seemingly paradoxical result naturally emerges out of an artificial neural network model of perceptual learning based on the incremental differentiation of stimulus representations: the self-organizing map (SOM). Like human listeners, networks show a sweet spot for easy-to-hard effects at intermediate levels of easy contrast somewhere between too easy and too hard. This trend is apparent across a wide range of free parameters. Analysis of network learning dynamics shows that easy-to-hard effects manifest when competition for network representational space is reduced (relative to constant hard training), but the learned stimuli are still similar enough to hard contrast stimuli to allow generalization. The data show that an incremental perceptual learning model can account for subtle characteristics of easy-to-hard effects where learning theories grounded in attentional discovery of relevant features cannot. We suggest several avenues for further development of the model in order to account for a wider range of auditory learning phenomena, its use in the optimization of auditory training regimens, and its role in elucidating perceptual learning mechanisms.

Motor adaptation is reduced by symbolic compared to sensory feedback.

Motor adaptation - the process of reducing motor errors through feedback and practice - is an essential feature of human competence, allowing us to move accurately in dynamic and novel environments. Adaptation typically results from sensory feedback, with most learning driven by visual and proprioceptive feedback that arises with the movement. In humans, motor adaptation can also be driven by symbolic feedback. In the present study, we examine how implicit and explicit components of motor adaptation are modulated by symbolic feedback. We conducted three reaching experiments involving over 400 human participants to compare sensory and symbolic feedback using a task in which both types of learning processes could be operative (Experiment 1) or tasks in which learning was expected to be limited to only an explicit process (Experiments 2 and 3). Adaptation with symbolic feedback was dominated by explicit strategy use, with minimal evidence of implicit recalibration. Even when matched in terms of information content, adaptation to rotational and mirror reversal perturbations was slower in response to symbolic feedback compared to sensory feedback. Our results suggest that the abstract and indirect nature of symbolic feedback disrupts strategic reasoning and/or refinement, deepening our understanding of how feedback type influences the mechanisms of sensorimotor learning.

Quantitative susceptibility mapping based basal ganglia segmentation via AGSeg: leveraging active gradient guiding mechanism in deep learning.

With better visual contrast and the ability for magnetic susceptibility quantification analysis, quantitative susceptibility mapping (QSM) has emerged as an important magnetic resonance imaging (MRI) method for basal ganglia studies. Precise segmentation of basal ganglia is a prerequisite for quantification analysis of tissue magnetic susceptibility, which is crucial for subsequent disease diagnosis and surgical planning. The conventional method of localizing and segmenting basal ganglia heavily relies on layer-by-layer manual annotation by experts, resulting in a tedious amount of workload. Although several morphology registration and deep learning based methods have been developed to automate segmentation, the voxels around the nuclei boundary remain a challenge to distinguish due to insufficient tissue contrast. This paper proposes AGSeg, an active gradient guidance-based susceptibility and magnitude information complete (MIC) network for real-time and accurate basal ganglia segmentation. Various datasets, including clinical scans and data from healthy volunteers, were collected across multiple centers with different magnetic field strengths (3T/5T/7T), with a total of 210 three-dimensional (3D) susceptibility measurements. Manual segmentations following fixed rules for anatomical borders annotated by experts were used as ground truth labels. The proposed network took QSM maps and Magnitude images as two individual inputs, of which the features are selectively enhanced in the proposed magnitude information complete (MIC) module. AGSeg utilized a dual-branch architecture, with Seg-branch aiming to generate a proper segmentation map and Grad-branch to reconstruct the gradient map of regions of interest (ROIs). With the support of the newly designed active gradient module (AGM) and gradient guiding module (GGM), the Grad-branch provided attention guidance for the Seg-branch, facilitating it to focus on the boundary of target nuclei. Ablation studies were conducted to assess the functionality of the proposed modules. Significant performance decrement was observed after ablating relative modules. AGSeg was evaluated against several existing methods on both healthy and clinical data, achieving an average Dice similarity coefficient (DSC) =0.874 and average 95% Hausdorff distance (HD95) =2.009. Comparison experiments indicated that our model had superior performance on basal ganglia segmentation and better generalization ability over existing methods. The AGSeg outperformed all implemented comparison deep learning algorithms with average DSC enhancement ranging from 0.036 to 0.074. The current work integrates a deep learning-based method into automated basal ganglia segmentation. The high processing speed and segmentation robustness of AGSeg contribute to the feasibility of future surgery planning and intraoperative navigation. Experiments show that leveraging active gradient guidance mechanisms and magnitude information completion can facilitate the segmentation process. Moreover, this approach also offers a portable solution for other multi-modality medical image segmentation tasks.

Developing Deep LSTMs With Later Temporal Attention for Predicting COVID-19 Severity, Clinical Outcome, and Antibody Level by Screening Serological Indicators Over Time.

The clinical course of COVID-19, as well as the immunological reaction, is notable for its extreme variability. Identifying the main associated factors might help understand the disease progression and physiological status of COVID-19 patients. The dynamic changes of the antibody against Spike protein are crucial for understanding the immune response. This work explores a temporal attention (TA) mechanism of deep learning to predict COVID-19 disease severity, clinical outcomes, and Spike antibody levels by screening serological indicators over time. We use feature selection techniques to filter feature subsets that are highly correlated with the target. The specific deep Long Short-Term Memory (LSTM) models are employed to capture the dynamic changes of disease severity, clinical outcome, and Spike antibody level. We also propose deep LSTMs with a TA mechanism to emphasize the later blood test records because later records often attract more attention from doctors. Risk factors highly correlated with COVID-19 are revealed. LSTM achieves the highest classification accuracy for disease severity prediction. Temporal Attention Long Short-Term Memory (TA-LSTM) achieves the best performance for clinical outcome prediction. For Spike antibody level prediction, LSTM achieves the best permanence. The experimental results demonstrate the effectiveness of the proposed models. The proposed models can provide a computer-aided medical diagnostics system by simply using time series of serological indicators.

Deep learning-based channel estimation with application to 5G and beyond networks

Channel state information (CSI) feedback estimation for a downlink medium in a massive multiple input multiple output (MIMO) system is an essential and critical task to improve channel capacity and performance yield, especially in a frequency division duplex (FDD) multiplexing system. However, spectral efficiency degradation is a massive issue due to high channel feedback overhead. This work proposes a deep learning-based channel estimation (DLCE) model to improve channel reconstruction efficiency and channel overhead reduction accuracy. The proposed deep learning (DL) mechanism consists of encoder and decoder network where encoder network is utilized to compress CSI matrices whereas decoder network is used to decompress obtained CSI matrices. Here, inverse discrete Fourier transform (IDFT) method is utilized to convert CSI matrices of frequency domain into CSI matrices of delay domain. Simulation results are evaluated between uplink and downlink medium in the massive MIMO system considering a co-operation in science and technology (COST) 2,100 model. Here, a significant improvement in correlation and normalized mean square error (NMSE) results is observed. The proposed DLCE model shows superior performance against varied channel estimation techniques in terms of NMSE and correlation efficiency.