Amount Of Training Data Research Articles

End-to-end adaptive object detection with learnable Retinex for low-light city environment

ABSTRACT In the smart city context, efficient urban surveillance under low-light conditions is crucial. Accurate object detection in dimly lit areas is vital for safety and nighttime driving. However, subpar, poorly lit images due to environmental or equipment limitations pose a challenge, affecting precision in tasks like object detection and segmentation. Existing solutions often involve time-consuming, inefficient image preprocessing and lack strong theoretical support for low-light city image enhancement. To address these issue, we propose an end-to-end pipeline named LAR-YOLO that leverages convolutional network to extract a set of image transformation parameters, and implements the Retinex theory to proficiently elevate the quality of the image. Unlike conventional approaches, this innovative method eliminates the need for hand-crafted parameters and can adaptively enhance each low-light image. Additionally, due to a restricted quantity of training data, the detection model may not achieve an adequate level of expertise to enhance detection accuracy. To tackle this challenge, we introduce a cross-domain learning approach that supplements the low-light model with knowledge from normal light scenarios. Our proof-of-principle experiments and ablation studies utilising ExDark and VOC datasets demonstrate that our proposed method outperforms similar low-light object detection algorithms by approximately 13% in terms of accuracy.

Exploiting Cultural Biases via Homoglyphs in Text-to-Image Synthesis

Models for text-to-image synthesis, such as DALL-E 2 and Stable Diffusion, have recently drawn a lot of interest from academia and the general public. These models are capable of producing high-quality images that depict a variety of concepts and styles when conditioned on textual descriptions. However, these models adopt cultural characteristics associated with specific Unicode scripts from their vast amount of training data, which may not be immediately apparent. We show that by simply inserting single non-Latin characters in the textual description, common models reflect cultural biases in their generated images. We analyze this behavior both qualitatively and quantitatively and identify a model’s text encoder as the root cause of the phenomenon. Such behavior can be interpreted as a model feature, offering users a simple way to customize the image generation and reflect their own cultural background. Yet, malicious users or service providers may also try to intentionally bias the image generation. One goal might be to create racist stereotypes by replacing Latin characters with similarly-looking characters from non-Latin scripts, so-called homoglyphs. To mitigate such unnoticed script attacks, we propose a novel homoglyph unlearning method to fine-tune a text encoder, making it robust against homoglyph manipulations.

A Few-Shot-Based Model-Agnostic Meta-Learning for Intrusion Detection in Security of Internet of Things

Currently, the effective technologies used to protect the security of the Internet of Things include blockchain and intrusion detection systems. The traditional IoT intrusion detection systems based on supervised learning usually requires a large amount of data for training, with high costs, and most of them can only work in several specific types of attacks. Faced with the variability of current IoT attack methods and the complexity of the network environment, they are usually unable to play a role in a short time. Especially in dealing with new security problems such as Zero-day attacks, which have a small number of learnable samples. Therefore, this paper proposed an IoT intrusion detection model to deal with the situation where learnable samples are insufficient. The model adopts the idea of meta-learning and divides the training process into two layers based on model-agnostic meta-learning. Its purpose is to train a relatively general model using the malicious network flow data of various known attack modes. To this end, a few-shot IoT intrusion detection dataset, FSIDS-IoT, is constructed based on five datasets, namely CIC-DDoS2019, CIC-IDS2017, CSE-CIC-IDS2018, NSL-KDD, and UNSW-NB15. Experiments show that our proposed approach, using only a few gradient steps and a small amount of training data from the new attack type, can perform well to protect cyber security. In the test of identifying unknown attacks, the optimal accuracy under the 5-way 1-shot setting reached 78.26%, 90.09% under the 5-way 5-shot setting, and 92.19% under the 5-way 10-shot setting.

Analysis and Improvement of Channel Equalization for Underwater Acoustic Communication Based on Variational Mode Decomposition and CNN-BiLSTM

Neural networks, including conventional neural networks (CNNs) and long- and short-term memory (LSTM) networks, have gained considerable attention in recent years due to their remarkable ability to recognize signals, which make them an ideal option for communication applications in aquatic environments. Despite the promising results achieved through the integration of various algorithms and CNN-LSTM networks, the utilization of these networks necessitates a significant amount of data for training which results in increased complexity of recognition systems. Therefore, it is crucial to explore new solutions that can mitigate these issues while also enhancing the overall performance of underwater acoustic communication systems. Therefore, this paper proposes a VCB framework that integrates the variational modal decomposition (VMD) algorithm, CNN and Bidirectional LSTM (BiLSTM) networks. This framework employs the VMD algorithm to extract energy characteristics from signals that are modulated by filter bank multi-carrier (FBMC) technologies. One-dimensional convolution (Conv1D) is then employed to extract the abstract features, which are subsequently fed into LSTM for time correlation analysis, from the energy values at each time step. Experimental results indicate that the neural network framework utilized in this study significantly diminishes data volume and enhances the recognition rate of FBMC signals, resulting in a lower bit error rate (BER).

Improving snore detection under limited dataset through harmonic/percussive source separation and convolutional neural networks

Snoring, an acoustic biomarker commonly observed in individuals with Obstructive Sleep Apnoea Syndrome (OSAS), holds significant potential for diagnosing and monitoring this recognized clinical disorder. Irrespective of snoring types, most snoring instances exhibit identifiable harmonic patterns manifested through distinctive energy distributions over time. In this work, we propose a novel method to differentiate monaural snoring from non-snoring sounds by analyzing the harmonic content of the input sound using harmonic/percussive sound source separation (HPSS). The resulting feature, based on the harmonic spectrogram from HPSS, is employed as input data for conventional neural network architectures, aiming to enhance snoring detection performance even under a limited data learning framework. To evaluate the performance of our proposal, we studied two different scenarios: 1) using a large dataset of snoring and interfering sounds, and 2) using a reduced training set composed of around 1% of the data material. In the former scenario, the proposed HPSS-based feature provides competitive results compared to other input features from the literature. However, the key advantage of the proposed method lies in the superior performance of the harmonic spectrogram derived from HPSS in a limited data learning context. In this particular scenario, using the proposed harmonic feature significantly enhances the performance of all the studied architectures in comparison to the classical input features documented in the existing literature. This finding clearly demonstrates that incorporating harmonic content enables more reliable learning of the essential time-frequency characteristics that are prevalent in most snoring sounds, even in scenarios where the amount of training data is limited.

Stroke classification and treatment support system artificial intelligence for usefulness of stroke diagnosis.

It is important to diagnose cerebral infarction at an early stage and select an appropriate treatment method. The number of stroke-trained physicians is unevenly distributed; thus, a shortage of specialists is a major problem in some regions. In this retrospective design study, we tested whether an artificial intelligence (AI) we built using computer-aided detection/diagnosis may help medical physicians to classify stroke for the appropriate treatment. To build the Stroke Classification and Treatment Support System AI, the clinical data of 231 hospitalized patients with ischemic stroke from January 2016 to December 2017 were used for training the AI. To verify the diagnostic accuracy, 151 patients who were admitted for stroke between January 2018 and December 2018 were also enrolled. By utilizing multimodal data, such as DWI and ADC map images, as well as patient examination data, we were able to construct an AI that can explain the analysis results with a small amount of training data. Furthermore, the AI was able to classify with high accuracy (Cohort 1, evaluation data 88.7%; Cohort 2, validation data 86.1%). In recent years, the treatment options for cerebral infarction have increased in number and complexity, making it even more important to provide appropriate treatment according to the initial diagnosis. This system could be used for initial treatment to automatically diagnose and classify strokes in hospitals where stroke-trained physicians are not available and improve the prognosis of cerebral infarction.

Sustainable Coffee Leaf Diagnosis: A Deep Knowledgeable Meta-Learning Approach

Multi-task visual recognition plays a pivotal role in addressing the composite challenges encountered during the monitoring of crop health, pest infestations, and disease outbreaks in precision agriculture. Machine learning approaches have been revolutionizing the diagnosis of plant disease in recent years; however, they require a large amount of training data and suffer from limited generalizability for unseen data. This work introduces a novel knowledgeable meta-learning framework for the few-shot multi-task diagnosis of biotic stress in coffee leaves. A mixed vision transformer (MVT) learner is presented to generate mixed contextual attention maps from discriminatory latent representations between support and query images to give more emphasis to the biotic stress lesions in coffee leaves. Then, a knowledge distillation strategy is introduced to avoid disastrous forgetting phenomena during inner-loop training. An adaptive meta-training rule is designed to automatically update the parameters of the meta-learner according to the current task. The competitive results from exhaustive experimentations on public datasets demonstrate the superior performance of our approach over the traditional methods. This is not only restricted to enhancing the accuracy and efficiency of coffee leaf disease diagnosis but also contributes to reducing the environmental footprint through optimizing resource utilization and minimizing the need for chemical treatments, hence aligning with broader sustainability goals in agriculture.

Experimental studies for improving the performance of children's speaker verification system using short utterances

The task of developing an automatic speaker verification (ASV) system for children's speech is a formidable one due to a number of reasons. The dearth of domain-specific data is one among them. The challenge further intensifies with the introduction of short utterances of speech, a relatively unexplored domain in the case of children's ASV. Voice-based biometric systems suffers miserably when speech data, inadequate both in volume as well as in duration, is used either for enrollment or verification. To circumvent the issue arising due to data scarcity, the work in this paper extensively explores in-domain as well as various out-of-domain data augmentation techniques. A data augmentation approach is proposed that encompasses both in-domain and out-of-domain data augmentation techniques. The in-domain data augmentation approach, incorporates speed perturbation of children's speech. The out-of-domain data used are from adult speakers which are known to have acoustic attributes in stark contrast to child speakers. The acoustic characteristics of the adult speech data in this study are altered on two fronts namely speech waveform modification and feature-level modification, in order to modify the adult acoustic features and render it acoustically similar to children's speech prior to augmentation. While the speech waveform modification involves various signal processing techniques like prosody modification, formant modification and voice-conversion. The feature-level modification on the other hand involves Vocal-tract length normalization technique (VTLN) which explicitly models and compensates for the ill-effects of variations in vocal tract length by linearly warping the frequency axis of speech signals. The proposed data augmentation approach helps not only in increasing the amount of training data but also in effectively capturing the missing target attributes which helps in boosting the verification performance. A relative improvement of 48.01% in equal error rate (EER) with respect to the baseline system is a testimony of it. Furthermore, the conventionally used Mel-frequency cepstral coefficients (MFCC) are known to average out the higher-frequency components. Prior literary works have shown that a significant amount of relevant acoustic information is available in the higher-frequency region of the children's speech. Therefore, effective preservation of higher-frequency contents in children's speech is of paramount importance which must be appropriately tackled for the development of a reliable and robust children's ASV system. In this regard, frame-level concatenation of the MFCC features with the Inverse-Mel-frequency cepstral coefficient (IMFCC) features is undertaken. The feature concatenation of MFCC and IMFCC is carried out with the sole intention of effectively preserving the higher-frequency contents in the children's speech data. The low canonical correlation existing between the MFCC and the IMFCC feature vectors provides the necessary impetus to go with their feature fusion. The feature concatenation approach, when combined with proposed data augmentation, helps in further improvement of the verification performance. The experimental results testify our claims, wherein we have achieved an overall relative reduction of 50.15% for equal error rate.

Feed-forward networks using logistic regression and support vector machine for whole-slide breast cancer histopathology image classification

The performance of an image classification depends on the efficiency of the feature learning process. This process is a challenging task that traditionally requires prior knowledge from domain experts. Recently, representation learning was introduced to extract features directly from the raw images without any prior knowledge. Deep learning using a Convolutional Neural Network (CNN) has gained massive attention for performing image classification, as it achieves remarkable accuracy that sometimes exceeds human performance. But this type of network learns features by using a back-propagation approach. This approach requires a huge amount of training data and suffers from the vanishing gradient problem that deteriorates the feature learning. The forward-propagation approach uses predefined filters or filters learned outside the model and applied in a feed-forward manner. This approach is proven to achieve good results with small size labeled datasets. In this work, we investigate the suitability of using two feed-forward methods such as Convolutional Logistic Regression Network (CLR), and Convolutional Support Vector Machine Network for Histopathology Images (CSVM-H). The experiments we have conducted on two small breast cancer datasets (Sultan Qaboos University Hospital (SQUH) and BreaKHis dataset) demonstrate the advantage of using feed-forward approaches over the traditional back-propagation ones. On those datasets, the proposed models CLR and CSVM-H were faster to train and achieved better classification performance than the traditional back-propagation methods (VggNet-16 and ResNet-50) on the SQUH dataset. Importantly, our proposed approach CLR and CSVM-H efficiently learn representations from small amounts of breast cancer whole-slide images and achieve an AUC of 0.83 and 0.84, respectively, on the SQUH dataset. Moreover, the proposed models reduce memory footprint in the classification of Whole-Slide histopathology images since their training time is significantly reduced compared to the traditional CNN on the SQUH and BreaKHis datasets.

Developing a Data-Driven Transfer Learning Model to Locate Tropical Cyclone Centers on Satellite Infrared Imagery

Abstract In this paper, a data-driven transfer learning (TL) model for locating tropical cyclone (TC) centers from satellite infrared images in the northwest Pacific is developed. A total of 2450 satellite infrared TC images derived from 97 TCs between 2015 and 2018 were used for this paper. The TC center location model (ResNet-TCL) with added residual fully connected modules is built for the TC center location. The MAE of the ResNet-TCL model is 34.8 km. Then TL is used to improve the model performance, including obtaining a pretrained model based on the ImageNet dataset, transferring the pretrained model parameters to the ResNet-TCL model, and using TC satellite infrared imagery to fine-train the ResNet-TCL model. The results show that the TL-based model improves the location accuracy by 14.1% (29.3 km) over the no-TL model. The model performance increases logarithmically with the amount of training data. When the training data are large, the benefit of increasing the training samples is smaller than the benefit of using TL. The comparison of model results with the best track data of TCs shows that the MAEs of TCs center is 29.3 km for all samples and less than 20 km for H2–H5 TCs. In addition, the visualization of the TL-based TC center location model shows that the TL model can accurately extract the most important features related to TC center location, including TC eye, TC texture, and contour. On the other hand, the no-TL model does not accurately extract these features.

One-Vote Veto: Semi-Supervised Learning for Low-Shot Glaucoma Diagnosis.

Convolutional neural networks (CNNs) are a promising technique for automated glaucoma diagnosis from images of the fundus, and these images are routinely acquired as part of an ophthalmic exam. Nevertheless, CNNs typically require a large amount of well-labeled data for training, which may not be available in many biomedical image classification applications, especially when diseases are rare and where labeling by experts is costly. This article makes two contributions to address this issue: 1) It extends the conventional Siamese network and introduces a training method for low-shot learning when labeled data are limited and imbalanced, and 2) it introduces a novel semi-supervised learning strategy that uses additional unlabeled training data to achieve greater accuracy. Our proposed multi-task Siamese network (MTSN) can employ any backbone CNN, and we demonstrate with four backbone CNNs that its accuracy with limited training data approaches the accuracy of backbone CNNs trained with a dataset that is 50 times larger. We also introduce One-Vote Veto (OVV) self-training, a semi-supervised learning strategy that is designed specifically for MTSNs. By taking both self-predictions and contrastive predictions of the unlabeled training data into account, OVV self-training provides additional pseudo labels for fine-tuning a pre-trained MTSN. Using a large (imbalanced) dataset with 66,715 fundus photographs acquired over 15 years, extensive experimental results demonstrate the effectiveness of low-shot learning with MTSN and semi-supervised learning with OVV self-training. Three additional, smaller clinical datasets of fundus images acquired under different conditions (cameras, instruments, locations, populations) are used to demonstrate the generalizability of the proposed methods.

Subspace Model-Assisted Deep Learning for Improved Image Reconstruction.

Image reconstruction from limited and/or sparse data is known to be an ill-posed problem and a priori information/constraints have played an important role in solving the problem. Early constrained image reconstruction methods utilize image priors based on general image properties such as sparsity, low-rank structures, spatial support bound, etc. Recent deep learning-based reconstruction methods promise to produce even higher quality reconstructions by utilizing more specific image priors learned from training data. However, learning high-dimensional image priors requires huge amounts of training data that are currently not available in medical imaging applications. As a result, deep learning-based reconstructions often suffer from two known practical issues: a) sensitivity to data perturbations (e.g., changes in data sampling scheme), and b) limited generalization capability (e.g., biased reconstruction of lesions). This paper proposes a new method to address these issues. The proposed method synergistically integrates model-based and data-driven learning in three key components. The first component uses the linear vector space framework to capture global dependence of image features; the second exploits a deep network to learn the mapping from a linear vector space to a nonlinear manifold; the third is an unrolling-based deep network that captures local residual features with the aid of a sparsity model. The proposed method has been evaluated with magnetic resonance imaging data, demonstrating improved reconstruction in the presence of data perturbation and/or novel image features. The method may enhance the practical utility of deep learning-based image reconstruction.

Backdoor attack and defense in federated generative adversarial network-based medical image synthesis.

Deep Learning-based image synthesis techniques have been applied in healthcare research for generating medical images to support open research and augment medical datasets. Training generative adversarial neural networks (GANs) usually require large amounts of training data. Federated learning (FL) provides a way of training a central model using distributed data while keeping raw data locally. However, given that the FL server cannot access the raw data, it is vulnerable to backdoor attacks, an adversarial by poisoning training data. Most backdoor attack strategies focus on classification models and centralized domains. It is still an open question if the existing backdoor attacks can affect GAN training and, if so, how to defend against the attack in the FL setting. In this work, we investigate the overlooked issue of backdoor attacks in federated GANs (FedGANs). The success of this attack is subsequently determined to be the result of some local discriminators overfitting the poisoned data and corrupting the local GAN equilibrium, which then further contaminates other clients when averaging the generator's parameters and yields high generator loss. Therefore, we proposed FedDetect, an efficient and effective way of defending against the backdoor attack in the FL setting, which allows the server to detect the client's adversarial behavior based on their losses and block the malicious clients. Our extensive experiments on two medical datasets with different modalities demonstrate the backdoor attack on FedGANs can result in synthetic images with low fidelity. After detecting and suppressing the detected malicious clients using the proposed defense strategy, we show that FedGANs can synthesize high-quality medical datasets (with labels) for data augmentation to improve classification models' performance.

Bone mineral density estimation from a plain X-ray image by learning decomposition into projections of bone-segmented computed tomography.

Osteoporosis is a prevalent bone disease that causes fractures in fragile bones, leading to a decline in daily living activities. Dual-energy X-ray absorptiometry (DXA) and quantitative computed tomography (QCT) are highly accurate for diagnosing osteoporosis; however, these modalities require special equipment and scan protocols. To frequently monitor bone health, low-cost, low-dose, and ubiquitously available diagnostic methods are highly anticipated. In this study, we aim to perform bone mineral density (BMD) estimation from a plain X-ray image for opportunistic screening, which is potentially useful for early diagnosis. Existing methods have used multi-stage approaches consisting of extraction of the region of interest and simple regression to estimate BMD, which require a large amount of training data. Therefore, we propose an efficient method that learns decomposition into projections of bone-segmented QCT for BMD estimation under limited datasets. The proposed method achieved high accuracy in BMD estimation, where Pearson correlation coefficients of 0.880 and 0.920 were observed for DXA-measured BMD and QCT-measured BMD estimation tasks, respectively, and the root mean square of the coefficient of variation values were 3.27 to 3.79% for four measurements with different poses. Furthermore, we conducted extensive validation experiments, including multi-pose, uncalibrated-CT, and compression experiments toward actual application in routine clinical practice.

Prediction of Short-Shot Defects in Injection Molding by Transfer Learning

For a long time, the traditional injection molding industry has faced challenges in improving production efficiency and product quality. With advancements in Computer-Aided Engineering (CAE) technology, many factors that could lead to product defects have been eliminated, reducing the costs associated with trial runs during the manufacturing process. However, despite the progress made in CAE simulation results, there still exists a slight deviation from actual conditions. Therefore, relying solely on CAE simulations cannot entirely prevent product defects, and businesses still need to implement real-time quality checks during the production process. In this study, we developed a Back Propagation Neural Network (BPNN) model to predict the occurrence of short-shots defects in the injection molding process using various process states as inputs. We developed a Back Propagation Neural Network (BPNN) model that takes injection molding process states as input to predict the occurrence of short-shot defects during the injection molding process. Additionally, we investigated the effectiveness of two different transfer learning methods. The first method involved training the neural network model using CAE simulation data for products with length–thickness ratios (LT) of 60 and then applying transfer learning with real process data. The second method trained the neural network model using real process data for products with LT60 and then applied transfer learning with real process data from products with LT100. From the results, we have inferred that transfer learning, as compared to conventional neural network training methods, can prevent overfitting with the same amount of training data. The short-shot prediction models trained using transfer learning achieved accuracies of 90.2% and 94.4% on the validation datasets of products with LT60 and LT100, respectively. Through integration with the injection molding machine, this enables production personnel to determine whether a product will experience a short-shot before the mold is opened, thereby increasing troubleshooting time.

Objective satellite methods including AI algorithms reviewed for the tenth International workshop on tropical cyclones (IWTC-10)

Here we explore the latest four years (2019–2022) of using satellite data to objectively analyze tropical cyclones (TC) and issue recommendations for improved analysis. We first discuss new methods of direct retrieval from SAR and geostationary imagers. Next, we survey some of the most prominent new techniques in AI and discuss their major capabilities (especially accuracy in nonlinear TC behavior, characterization of model uncertainty and creation of synthetic satellite imagery) and limitations (especially lack of transparency and limited amount of training data). We also identify concerns with biases and unlabeled uncertainties in the Best Track records as being a first-order limitation for further progress in objective methods. The article concludes with recommendations to improve future objective methods, especially in the area of more accurate and reliable training data sets.

Training Sample Pattern Optimization Based on a Swarm Intelligence Algorithm for Tiltrotor Flight Dynamics Model Approximation

Neural networks have been widely used as compensational models for aircraft control designs and as surrogate models for other optimizations. In the case of tiltrotor aircraft, the total number of aircraft states and controls is much greater than that of both traditional fixed-wings and helicopters. This requires, in general, a huge amount of training data for the network to reach a satisfactory approximation precision and makes the network size rise considerably. To solve the practical problem of reducing the size of the approximating network, efforts have to be made in the efficient utilization of the limited amount of training data. This work presents the methodology of optimizing the sample pattern of the training data set by adopting the metaheuristic algorithm of the particle swarm optimizer improved by the fourth-order Runge–Kutta algorithm. A 6-degree-of-freedom nonlinear flight dynamics model of the tiltrotor aircraft is derived, along with its approximation radial basis function neural network. An example case of approximating a highly nonlinear function is studied to illustrate the principle and main parameters of the optimizer, and the approximation performance of the time-domain response of the unstable nonlinear system is revealed by the study of a Van der Pol oscillator. Then, the presented method is applied to the modeled tiltrotor aircraft for its early and late conversion modes. The optimization scheme shows great improvement in both cases, as the function approximation error is reduced significantly.

Federated Learning: Centralized and P2P for a Siamese Deep Learning Model for Diabetes Foot Ulcer Classification

It is a known fact that AI models need massive amounts of data for training. In the medical field, the data are not necessarily available at a single site but are distributed over several sites. In the field of medical data sharing, particularly among healthcare institutions, the need to maintain the confidentiality of sensitive information often restricts the comprehensive utilization of real-world data in machine learning. To address this challenge, our study experiments with an innovative approach using federated learning to enable collaborative model training without compromising data confidentiality and privacy. We present an adaptation of the federated averaging algorithm, a predominant centralized learning algorithm, to a peer-to-peer federated learning environment. This adaptation led to the development of two extended algorithms: Federated Averaging Peer-to-Peer and Federated Stochastic Gradient Descent Peer-to-Peer. These algorithms were applied to train deep neural network models for the detection and monitoring of diabetic foot ulcers, a critical health condition among diabetic patients. This study compares the performance of Federated Averaging Peer-to-Peer and Federated Stochastic Gradient Descent Peer-to-Peer with their centralized counterparts in terms of model convergence and communication costs. Additionally, we explore enhancements to these algorithms using targeted heuristics based on client identities and f1-scores for each class. The results indicate that models utilizing peer-to-peer federated averaging achieve a level of convergence that is comparable to that of models trained via conventional centralized federated learning approaches. This represents a notable progression in the field of ensuring the confidentiality and privacy of medical data for training machine learning models.

A Vision-Based Pose Estimation of a Non-Cooperative Target Based on a Self-Supervised Transformer Network

In the realm of non-cooperative space security and on-orbit service, a significant challenge is accurately determining the pose of abandoned satellites using imaging sensors. Traditional methods for estimating the position of the target encounter problems with stray light interference in space, leading to inaccurate results. Conversely, deep learning techniques require a substantial amount of training data, which is especially difficult to obtain for on-orbit satellites. To address these issues, this paper introduces an innovative binocular pose estimation model based on a Self-supervised Transformer Network (STN) to achieve precise pose estimation for targets even under poor imaging conditions. The proposed method generated simulated training samples considering various imaging conditions. Then, by combining the concepts of convolutional neural networks (CNN) and SIFT features for each sample, the proposed method minimized the disruptive effects of stray light. Furthermore, the feedforward network in the Transformer employed in the proposed method was replaced with a global average pooling layer. This integration of CNN’s bias capabilities compensates for the limitations of the Transformer in scenarios with limited data. Comparative analysis against existing pose estimation methods highlights the superior robustness of the proposed method against variations caused by noisy sample sets. The effectiveness of the algorithm is demonstrated through simulated data, enhancing the current landscape of binocular pose estimation technology for non-cooperative targets in space.

Deep Random Forest and AraBert for Hate Speech Detection from Arabic Tweets

Nowadays, hate speech detection from Arabic tweets attracts the attention of many researchers. Numerous systems and techniques have been proposed to address this classification challenge. Nonetheless, three major limits persist: the use of deep learning models with an excess of hyperparameters, the reliance on hand-crafted features, and the requirement for a huge amount of training data to achieve satisfactory performance. In this study, we propose Contextual Deep Random Forest (CDRF), a hate speech detection approach that combines contextual embedding and Deep Random Forest. From the experimental findings, the Arabic contextual embedding model proves to be highly effective in hate speech detection, outperforming the static embedding models. Additionally, we prove that the proposed CDRF significantly enhances the performance of Arabic hate speech classification.