Search Results (786)

The current process of embryo selection in in vitro fertilization is based on morphological criteria; embryos are manually evaluated by embryologists under subjective assessment. In this study, a deep learning-based pipeline was developed to classify the viability of embryos using combined inputs, including microscopic images of embryos and additional features, such as patient age and developed pseudo-features, including a continuous interpretation of Istanbul grading scores by predicting the embryo stage, inner cell mass, and trophectoderm. For viability prediction, convolution-based transferred learning models were employed, multiple pretrained models were compared, and image preprocessing techniques and hyperparameter optimization via Optuna were utilized. In addition, a custom weight was trained using a self-supervised learning framework known as the Simple Framework for Contrastive Learning of Visual Representations (SimCLR) in cooperation with generated images using generative adversarial networks (GANs). The best model was developed from the EfficientNet-B0 model using preprocessed images combined with pseudo-features generated using separate EfficientNet-B0 models, and optimized by Optuna to tune the hyperparameters of the models. The designed model’s F1 score, accuracy, sensitivity, and area under curve (AUC) were 65.02%, 69.04%, 56.76%, and 66.98%, respectively. This study also showed an advantage in accuracy and a similar AUC when compared with the recent ensemble method. Full article

Change detection (CD) is an important research direction in the field of remote sensing, which aims to analyze the changes in the same area over different periods and is widely used in urban planning and environmental protection. While supervised learning methods in change detection have demonstrated substantial efficacy, they are often hindered by the rising costs associated with data annotation. Semi-supervised methods have attracted increasing interest, offering promising results with limited data labeling. These approaches typically employ strategies such as consistency regularization, pseudo-labeling, and generative adversarial networks. However, they usually face the problems of insufficient data augmentation and unbalanced quality and quantity of pseudo-labeling. To address the above problems, we propose a semi-supervised change detection method with data augmentation and adaptive threshold updating (DA-AT) for high-resolution remote sensing images. Firstly, a channel-level data augmentation (CLDA) technique is designed to enhance the strong augmentation effect and improve consistency regularization so as to address the problem of insufficient feature representation. Secondly, an adaptive threshold (AT) is proposed to dynamically adjust the threshold during the training process to balance the quality and quantity of pseudo-labeling so as to optimize the self-training process. Finally, an adaptive class weight (ACW) mechanism is proposed to alleviate the impact of the imbalance between the changed classes and the unchanged classes, which effectively enhances the learning ability of the model for the changed classes. We verify the effectiveness and robustness of the proposed method on two high-resolution remote sensing image datasets, WHU-CD and LEVIR-CD. We compare our method to five state-of-the-art change detection methods and show that it achieves better or comparable results. Full article

Recent improvements in self-driving and connected cars promise to enhance traffic safety by reducing risks and accidents. However, security concerns limit their acceptance. These vehicles, interconnected with infrastructure and other cars, are vulnerable to cyberattacks, which could lead to severe costs, including physical injury or death. In this article, we propose a framework for an intrusion detection system to protect internal vehicle communications from potential attacks and ensure secure sent/transferred data. In the proposed system, real auto-network datasets with Spoofing, DoS, and Fuzzy attacks are used. To accurately distinguish between benign and malicious messages, this study employed seven distinct supervised machine-learning algorithms for data classification. The selected algorithms encompassed Decision Trees, Random Forests, Naive Bayes, Logistic Regression, XG Boost, LightGBM, and Multi-layer Perceptrons. The proposed detection system performed well on large real-car hacking datasets. We achieved high accuracy in identifying diverse electronic intrusions across the complex internal networks of connected and autonomous vehicles. Random Forest and LightGBM outperformed the other algorithms examined. Random Forest outperformed the other algorithms in the merged dataset trial, with 99.9% accuracy and the lowest computing cost. The LightGBM algorithm, on the other hand, performed admirably in the domain of binary classification, obtaining the same remarkable 99.9% accuracy with no computing overhead. Full article

Fourier ptychographic microscopy (FPM) has recently emerged as an important non-invasive imaging technique which is capable of simultaneously achieving high resolution, wide field of view, and quantitative phase imaging. However, FPM still faces challenges in the image reconstruction due to factors such as noise, optical aberration, and phase wrapping. In this work, we propose a semi-supervised Fourier ptychographic transformer network (SFPT) for improved image reconstruction, which employs a two-stage training approach to enhance the image quality. First, self-supervised learning guided by low-resolution amplitudes and Zernike modes is utilized to recover pupil function. Second, a supervised learning framework with augmented training datasets is applied to further refine reconstruction quality. Moreover, the unwrapped phase is recovered by adjusting the phase distribution range in the augmented training datasets. The effectiveness of the proposed method is validated by using both the simulation and experimental data. This deep-learning-based method has potential applications for imaging thicker biology samples. Full article

Two challenges in computer vision (CV) related to face detection are the difficulty of acquisition in the target domain and the degradation of image quality. Especially in low-light situations, the poor visibility of images is difficult to label, which results in detectors trained under well-lit conditions exhibiting reduced performance in low-light environments. Conventional works image enhancement and object detection techniques are unable to resolve the inherent difficulties in collecting and labeling low-light images. The Dark-Illuminated Network with Contrastive Language–Image Pretraining (CLIP) and Self-Supervised Vision Transformer (Dino), abbreviated as DAl-CLIP-Dino is proposed to address the degradation of object detection performance in low-light environments and achieve zero-shot day–night domain adaptation. Specifically, an advanced reflectance representation learning module (which leverages Retinex decomposition to extract reflectance and illumination features from both low-light and well-lit images) and an interchange–redecomposition coherence process (which performs a second decomposition on reconstructed images after the exchange to generate a second round of reflectance and illumination predictions while validating their consistency using redecomposition consistency loss) are employed to achieve illumination invariance and enhance model performance. CLIP (VIT-based image encoder part) and Dino have been integrated for feature extraction, improving performance under extreme lighting conditions and enhancing its generalization capability. Our model achieves a mean average precision (mAP) of 29.6% for face detection on the DARK FACE dataset, outperforming other models in zero-shot domain adaptation for face detection. Full article

The demand for flexible automation in manufacturing has increased, incorporating vision-guided systems for object grasping. However, a key challenge is in-hand error, where discrepancies between the actual and estimated positions of an object in the robot’s gripper impact not only the grasp but also subsequent assembly stages. Corrective strategies used to compensate for misalignment can increase cycle times or rely on pre-labeled datasets, offline training, and validation processes, delaying deployment and limiting adaptability in dynamic industrial environments. Our main contribution is an online self-supervised learning method that automates data collection, training, and evaluation in real time, eliminating the need for offline processes. Building on this, our system collects real-time data during each assembly cycle, using corrective strategies to adjust the data and autonomously labeling them via a self-supervised approach. It then builds and evaluates multiple regression models through an auto machine learning implementation. The system selects the best-performing model to correct the misalignment and dynamically chooses between corrective strategies and the learned model, optimizing the cycle times and improving the performance during the cycle, without halting the production process. Our experiments show a significant reduction in the cycle time while maintaining the performance. Full article

The upsampling of point clouds is a common task to increase the expressiveness and richness of the details. The quality of upsampled point clouds is crucial for downstream tasks, such as mesh reconstruction. With the rapid development of deep learning technology, many neural network-based methods have been proposed for point cloud upsampling. However, there are common challenges among these methods such as blurring sharper points (e.g., corner or edge points) and producing points gathered together. These problems are caused by similar feature replication or insufficient supervised information. To address these concerns, we present SSPU-FENet, a self-supervised network consisting of two modules specifically designed for geometric detail-preserved point cloud upsampling. The first module, called the feature enhancement module (FEM), aims to prevent feature blurring. This module retains important features such as edges and corners by using non-artificial encoding methods and learning mechanisms to avoid the creation of blurred points. The second module, called the 3D noise perturbation module (NPM), focuses on high-dimensional feature processing and addresses the challenges of feature similarity. This module adjusts the spacing of reconstructed points, ensuring that they are neither too close nor too far apart, thus maintaining point uniformity. In addition, SSPU-FENet proposes self-supervised loss functions that emphasize global shape consistency and local geometric structure consistency. These loss functions enable efficient network training, leading to superior upsampling results. Experimental results on various datasets show that the upsampling results of the SSPU-FENet are comparable to those of supervised learning methods and close to the ground truth (GT) point clouds. Furthermore, our evaluation metrics, such as the chamfer distance (CD, 0.0991), outperform the best methods (CD, 0.0998) in the case of 16× upsampling with 2048-point input. Full article

Current industrial robotic manipulators have made their lack of flexibility evident. The systems must know beforehand the piece and its position. To address this issue, contemporary approaches typically employ learning-based techniques, which rely on extensive amounts of data. To obtain vast data, an often sought tool is an extensive grasp dataset. This work introduces our Physics-Based Self-Supervised Grasp Pose Detection (PBSS-GPD) pipeline for model-based grasping point detection, which is useful for generating grasp pose datasets. Given a gripper-object pair, it samples grasping pose candidates using a modified version of GPD (implementing inner-grasps, CAD support…) and quantifies their quality using the MuJoCo physics engine and a grasp quality metric that takes into account the pose of the object over time. The system is optimized to run on CPU in headless-parallelized mode, with the option of running in a graphical interface or headless and storing videos of the process. The system has been validated obtaining grasping poses for a subset of Egad! objects using the Franka Panda two-finger gripper, compared with state-of-the-art grasp generation pipelines and tested in a real scenario. While our system achieves similar accuracy compared to a contemporary approach, 84% on the real-world validation, it has proven to be effective at generating grasps with good centering 18 times faster than the compared system. Full article

Recommendation systems offer an effective solution to information overload, finding widespread application across e-commerce, news platforms, and beyond. By analyzing interaction histories, these systems automatically filter and recommend items that are most likely to resonate with users. Recently, with the swift advancement of social networking, group recommendation has emerged as a compelling research area, enabling personalized recommendations for groups of users. Unlike individual recommendation, group recommendation must consider both individual preferences and group dynamics, thereby enhancing decision-making efficiency for groups. One of the key challenges facing recommendation algorithms is data sparsity, a limitation that is even more severe in group recommendation than in traditional recommendation tasks. While various group recommendation methods attempt to address this issue, many of them still rely on single-view modeling or fail to sufficiently account for individual user preferences within a group, limiting their effectiveness. This paper addresses the data sparsity issue to improve group recommendation performance, overcoming the limitations of overlooking individual user recommendation tasks and depending on single-view modeling. We propose MCSS (multi-view collaborative training and self-supervised learning), a novel framework that harnesses both multi-view collaborative training and self-supervised learning specifically for group recommendations. By incorporating both group and individual recommendation tasks, MCSS leverages graph convolution and attention mechanisms to generate three sets of embeddings, enhancing the model’s representational power. Additionally, we design self-supervised auxiliary tasks to maximize the data utility, further enhancing performance. Through multi-task joint training, the model generates refined recommendation lists tailored to each group and individual user. Extensive validation and comparison demonstrate the method’s robustness and effectiveness, underscoring the potential of MCSS to advance state-of-the-art group recommendation. Full article

Weibo sentiment analysis has gained prominence, particularly during the COVID-19 pandemic, as a means to monitor public emotions and detect emerging mental health trends. However, challenges arise from Weibo’s informal language, nuanced expressions, and stylistic features unique to social media, which complicate the accurate interpretation of sentiments. Existing models often fall short, relying on text-based methods that inadequately capture the rich emotional texture of Weibo posts, and are constrained by single loss functions that limit emotional depth. To address these limitations, we propose a novel framework incorporating a sentiment graph and self-supervised learning. Our approach introduces a “sentiment graph” that leverages both word-to-post and post-to-post relational connections, allowing the model to capture fine-grained sentiment cues and context-dependent meanings. Enhanced by a gated mechanism within the graph, our model selectively filters emotional signals based on intensity and relevance, improving its sensitivity to subtle variations such as sarcasm. Additionally, a self-supervised objective enables the model to generalize beyond labeled data, capturing latent emotional structures within the graph. Through this integration of sentiment graph and self-supervised learning, our approach advances Weibo sentiment analysis, offering a robust method for understanding the complex emotional landscape of social media. Full article

The early identification of forest wood-boring pests is essential for effective pest management. However, detecting infestation in the early stages is difficult, as larvae, such as the emerald ash borer (EAB), Agrilus planipennis Fairmaire (Coleoptera: Buprestidae), usually feed inside the trees. Acoustic sensors can detect the pulse signals generated by larval feeding or movement, but these sounds are often weak and easily masked by background noise. To address this, we propose a dual-branch time-frequency multi-dilated dense network (DBMDNet) for noise reduction. Our model decouples two denoising training objectives: a magnitude masking decoder for coarse denoising and a complex spectral decoder for further magnitude repair and phase correction. Additionally, to enhance global time-frequency modeling, we use three different multi-dilated dense blocks to effectively separate clean signals from noisy data. Given the difficult acquisition of clean larval activity signals, we describe a self-supervised training procedure that utilizes only noisy larval activity signals directly collected from the wild, without the need for paired clean signals. Experimental results demonstrate that our proposed approach achieves the optimal performance on various evaluation metrics while requiring fewer parameters (only 98.62 k) compared to competitive models, achieving an average signal-to-noise ratio (SNR) improvement of 17.45 dB and a log-likelihood ratio (LLR) of 0.14. Furthermore, using the larval activity signals enhanced by DBMDNet, most of the noise is suppressed, and the accuracy of the recognition model is also significantly improved. Full article

Most existing methods for magnetic resonance imaging (MRI) reconstruction with deep learning use fully supervised training, which assumes that a fully sampled dataset with a high signal-to-noise ratio (SNR) is available for training. In many circumstances, however, such a dataset is highly impractical or even technically infeasible to acquire. Recently, a number of self-supervised methods for MRI reconstruction have been proposed, which use sub-sampled data only. However, the majority of such methods, such as Self-Supervised Learning via Data Undersampling (SSDU), are susceptible to reconstruction errors arising from noise in the measured data. In response, we propose Robust SSDU, which provably recovers clean images from noisy, sub-sampled training data by simultaneously estimating missing k-space samples and denoising the available samples. Robust SSDU trains the reconstruction network to map from a further noisy and sub-sampled version of the data to the original, singly noisy, and sub-sampled data and applies an additive Noisier2Noise correction term upon inference. We also present a related method, Noiser2Full, that recovers clean images when noisy, fully sampled data are available for training. Both proposed methods are applicable to any network architecture, are straightforward to implement, and have a similar computational cost to standard training. We evaluate our methods on the multi-coil fastMRI brain dataset with novel denoising-specific architecture and find that it performs competitively with a benchmark trained on clean, fully sampled data. Full article

Reinforcement Learning (RL) is increasingly being applied to complex decision-making tasks such as financial trading. However, designing effective reward functions remains a significant challenge. Traditional static reward functions often fail to adapt to dynamic environments, leading to inefficiencies in learning. This paper presents a novel approach, called Self-Rewarding Deep Reinforcement Learning (SRDRL), which integrates a self-rewarding network within the RL framework. The SRDRL mechanism operates in two primary phases: First, supervised learning techniques are used to learn from expert knowledge by employing advanced time-series feature extraction models, including TimesNet and WFTNet. This step refines the self-rewarding network parameters by comparing predicted rewards with expert-labeled rewards, which are based on metrics such as Min-Max, Sharpe Ratio, and Return. In the second phase, the model selects the higher value between the expert-labeled and predicted rewards as the RL reward, storing it in the replay buffer. This combination of expert knowledge and predicted rewards enhances the performance of trading strategies. The proposed implementation, called Self-Rewarding Double DQN (SRDDQN), demonstrates that the self-rewarding mechanism improves learning and optimizes trading decisions. Experiments conducted on datasets including DJI, IXIC, and SP500 show that SRDDQN achieves a cumulative return of 1124.23% on the IXIC dataset, significantly outperforming the next best method, Fire (DQN-HER), which achieved 51.87%. SRDDQN also enhances the stability and efficiency of trading strategies, providing notable improvements over traditional RL methods. The integration of a self-rewarding mechanism within RL addresses a critical limitation in reward function design and offers a scalable, adaptable solution for complex, dynamic trading environments. Full article

Transformer-based time series forecasting models use patch tokens for temporal patterns and variate tokens to learn covariates’ dependencies. While patch tokens inherently facilitate self-supervised learning, variate tokens are more suitable for linear forecasters as they help to mitigate distribution drift. However, the use of variate tokens prohibits masked model pretraining, as masking an entire series is absurd. To close this gap, we propose LSPatch-T (Long–Short Patch Transfer), a framework that transfers knowledge from short-length patch tokens into full-length variate tokens. A key implementation is that we selectively transfer a portion of the Transformer encoder to ensure the linear design of the downstream model. Additionally, we introduce a robust frequency loss to maintain consistency across different temporal ranges. The experimental results show that our approach outperforms Transformer-based baselines (Transformer, Informer, Crossformer, Autoformer, PatchTST, iTransformer) on three public datasets (ETT, Exchange, Weather), which is a promising step forward in generalizing time series forecasting models. Full article

Advancements in Raman light sheet microscopy have provided a powerful, non-invasive, marker-free method for imaging complex 3D biological structures, such as cell cultures and spheroids. By combining 3D tomograms made by Rayleigh scattering, Raman scattering, and fluorescence detection, this modality captures complementary spatial and molecular data, critical for biomedical research, histology, and drug discovery. Despite its capabilities, Raman light sheet microscopy faces inherent limitations, including low signal intensity, high noise levels, and restricted spatial resolution, which impede the visualization of fine subcellular structures. Traditional enhancement techniques like Fourier transform filtering and spectral unmixing require extensive preprocessing and often introduce artifacts. More recently, deep learning techniques, which have shown great promise in enhancing image quality, face their own limitations. Specifically, conventional deep learning models require large quantities of high-quality, manually labeled training data for effective denoising and super-resolution tasks, which is challenging to obtain in multi-modal microscopy. In this study, we address these limitations by exploring advanced zero-shot and self-supervised learning approaches, such as ZS-DeconvNet, Noise2Noise, Noise2Void, Deep Image Prior (DIP), and Self2Self, which enhance image quality without the need for labeled and large datasets. This study offers a comparative evaluation of zero-shot and self-supervised learning methods, evaluating their effectiveness in denoising, resolution enhancement, and preserving biological structures in multi-modal Raman light sheet microscopic images. Our results demonstrate significant improvements in image clarity, offering a reliable solution for visualizing complex biological systems. These methods establish the way for future advancements in high-resolution imaging, with broad potential for enhancing biomedical research and discovery. Full article