A Survey on Deep Reinforcement Learning Algorithms for Robotic Manipulation

Sensors, Год журнала: 2023, Номер 23(7), С. 3762 - 3762

Опубликована: Апрель 5, 2023

Robotic manipulation challenges, such as grasping and object manipulation, have been tackled successfully with the help of deep reinforcement learning systems. We give an overview recent advances in algorithms for robotic tasks this review. begin by outlining fundamental ideas parts a system. The many algorithms, value-based methods, policy-based actor–critic approaches, that suggested are then covered. also examine numerous issues arisen when applying these to robotics tasks, well various solutions put forth deal issues. Finally, we highlight several unsolved research talk about possible future directions subject.

Язык: Английский

---

Machine learning-inspired intrusion detection system for IoT: Security issues and future challenges

Computers & Electrical Engineering, Год журнала: 2025, Номер 123, С. 110265 - 110265

Опубликована: Март 26, 2025

Язык: Английский

---

Carbon-based molecular properties efficiently predicted by deep learning-based quantum chemical simulation with large language models

Computers in Biology and Medicine, Год журнала: 2024, Номер 176, С. 108531 - 108531

Опубликована: Май 1, 2024

Язык: Английский

---

Solving partial differential equations using large-data models: a literature review

Artificial Intelligence Review, Год журнала: 2024, Номер 57(6)

Опубликована: Май 24, 2024

Abstract Mathematics lies at the heart of engineering science and is very important for capturing modeling diverse processes. These processes may be naturally-occurring or man-made. One problem in this regard advanced mathematical problems their analysis. Partial differential equations (PDEs) are useful tools to end. However, solving complex PDEs requires extensive computational resources techniques. Neural networks provide a way solve reliably. In regard, large-data models new generation techniques, which have large dependency capabilities. Hence, they can richly model accurately such PDEs. Some common include Convolutional neural (CNNs) derivatives, transformers, etc. literature survey, background introduced. A gentle introduction area using given. Various state-of-the-art discussed. Also, major issues future scope identified. Through it hoped that readers will gain an insight into pursue research interesting area.

Язык: Английский

---

Self-attention and long-range relationship capture network for underwater object detection

Journal of King Saud University - Computer and Information Sciences, Год журнала: 2024, Номер 36(2), С. 101971 - 101971

Опубликована: Фев. 1, 2024

Underwater object detection has been shown to exhibit significant potential for exploring underwater environments. However, datasets often suffer from degeneration due uneven light distribution, complex environment, and crowded dynamic background. Thus, performance would be degraded accordingly. In this paper, a large kernel convolutional network based on self-attention long-range relationship capture is proposed. Firstly, hybrid dilated attention mechanism proposed, which adopts the idea of convolution combines advantages self-attention. This can avoid defects while achieving adaptiveness relevance. Secondly, feature enhancement block called residual reconstructed module captures dependencies in extracts more critical contextual information, thus solving problem degradation accuracy degradation. Thirdly, an adaptive spatial fusion head constructed, directly learn how filter different features at layers spatially; useless information filtered out, only useful kept combination enhance capability further. Finally, proposed above three techniques. Extensive experiments were conducted well-known RUOD, Aquarium, URPC, MS COCO. Compared prior state-of-the-art methods, experimental findings demonstrate that approach obtains highest mAP 88.7%, 86.5%, 98.9%, 71.4%, respectively. represents improvement 1.2, 1.5, 8.5, 0.2 percentage, order. The model shows capacity function by applying local details, as well grasp global relationships, prioritize essential data, spatially irrelevant information.

Язык: Английский

---

Gene selection and cancer classification using interaction-based feature clustering and improved-binary Bat algorithm

Computers in Biology and Medicine, Год журнала: 2024, Номер 181, С. 109071 - 109071

Опубликована: Авг. 27, 2024

Язык: Английский

---

Robust watermarking algorithm for oblique photography 3D models based on multi-level curvature and weighted distance

Earth Science Informatics, Год журнала: 2025, Номер 18(3)

Опубликована: Март 11, 2025

Язык: Английский

---

Enhanced brain image security using a hybrid of lifting wavelet transform and support vector machine

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Март 20, 2025

Abstract Thanks to technological improvements, digital picture watermarking has emerged as a useful method for preventing unlawful use and manipulation of photographs. Providing robustness against geometrical assault while maintaining an adequate level security imperceptibility is basic challenge in watermarking. With the support vector machine (SVM) lifting wavelet transform (LWT), this study offers effective authentication approach image on medical images. To distinguish between region interest (ROI) non-region (NROI) image, SVM first employed article. After that, LWT used incorporate watermark data into image’s NROI section (cover image). Additionally, shared secret key been increase suggested scheme’s resilience. A vast database test method’s performance various scenarios. determine whether current plan was acceptable, examined several experimental investigations. The results give PSNR value 67.81 dB structural similarity index measure 0.9999, Where improvement percentage 13.9462 dB, showing durability proposed model.

Язык: Английский

---

Quadruped Robot System Based on Reinforcement Learning and Parallel Simulation Training

Communications in computer and information science, Год журнала: 2025, Номер unknown, С. 231 - 235

Опубликована: Янв. 1, 2025

Язык: Английский

---

A holistic securing approach to speech steganography system with message embedding and pause optimization

Multimedia Tools and Applications, Год журнала: 2025, Номер unknown

Опубликована: Апрель 22, 2025

Язык: Английский

---