The Experts below are selected from a list of 2646 Experts worldwide ranked by ideXlab platform
Mingyi Cui - One of the best experts on this subject based on the ideXlab platform.
-
EMEIT - A threshold denoising based floating Point Representation genetic algorithm
Proceedings of 2011 International Conference on Electronic & Mechanical Engineering and Information Technology, 2011Co-Authors: Mingyi CuiAbstract:Genetic algorithm (GA) was widely used to many engineering optimization fields. Encoding is one of difficult issues of GA research. Floating Point presentation (FPR) is of the advantage of higher precision and convenience of searching in great space. Noises were generated by the FPR in genetic operation environment. The noises have influence on the performance of GA. In this paper, the properties of the noises were mostly analyzed in inherit operation. A novel floating Point Representation genetic algorithm was proposed based wavelet threshold denoising mutation. Many experiments were made on it. The results of the research and the experiments indicate which the method is superior to other algorithms, is reliable in theory, and is feasible in technique.
-
Floating Point Representation Genetic Algorithm Based on Wavelet Decomposition
2010 2nd International Workshop on Intelligent Systems and Applications, 2010Co-Authors: Mingyi Cui, Cui WeiAbstract:Genetic algorithm (GA) is used widely to many fields. Coding is one of difficult issues of GA research. Floating Point presentation (FPR) is of the advantage of higher precision and convenience of searching in great space. FPR is superior to other codes in function optimization and restriction optimization. But, the noises were neglected by researches which were generated by FPR in genetic environment. This paper is based on wavelet decomposition, the noises are mapped to Haar basis, algorithm is made with denoising mutation, the algorithm is implemented by programming. The results of the research and the experiments indicate which the method is superior to other algorithms, is reliable in theory, and is feasible in technique.
-
Research on Genetic Algorithm of Floating Point Representation Denoising Mutation Based on DTCWT
2010 International Conference on Measuring Technology and Mechatronics Automation, 2010Co-Authors: Mingyi Cui, Cui WeiAbstract:Dual-tree complex wavelet transform (DTCWT) is of better properties than discrete wavelet transform (DWT). The attention of the researches was aroused by the noise and its influence on the performance of algorithm in genetic operation of floating Point Representation genetic algorithm (FPRGA). In order to eliminate the noise influence on algorithm properties, this paper proposes denoising mutation on floating Point Representation (FPR) with DTCWT in genetic operation. It presents a FPR denoising mutation genetic algorithm based on DTCWT (FDMGAD). It makes some experiments. The result of the research and the experiments indicates which FDMGAD is superior to other algorithms. The method is reliable in theory, is feasible in technique.
-
Orthonormal multiwavelet denoising mutation based FPRGA
2010 International Conference on Mechanic Automation and Control Engineering, 2010Co-Authors: Mingyi Cui, Nan SunAbstract:Encoding is one of difficult problems of research on genetic algorithm (GA). Floating Point Representation (FPR) is super to other encoding method in function and restriction optimization. But this problem was ignored by researchers how FPR was denoised in running environment of GA for enhancing the performance of GA. In this paper, it was proved by the wavelet theory that odd-length and even-length of chromosome encoded could all mutate with wavelet denoising. Based on the above research result, the floating Point Representation GA (FPRGA) with orthonormal multiwavelet denoising mutation (FGAMDM) was presented by the paper. The experiments were done by it. The result of the research and the experiment indicates that the method is credible in theory, is feasible in technique.
-
Research on floating Point Representation genetic algorithm based on wavelet threshold shrinkage denoising
2009 IEEE International Conference on Intelligent Computing and Intelligent Systems, 2009Co-Authors: Mingyi CuiAbstract:Floating Point Representation (FPR) is of the strongPoint of high precision and facilitating search on high-dimension space. It is superior to other Representation in function optimization and restriction optimization. But, the noise was brought about in run environment of floating Point Representation genetic algorithm (FPRGA). This was often neglected by researchers. Simple FPRGA uses bounded random mutation. It cannot avoid the noise to influence on the algorithm performance. This paper presents a floating Point Representation genetic algorithm based on wavelet threshold shrinkage denoising (FGAWSD). A filter was structured. Mutation operation was replaced with different thresholds denoising. The experiments were done. The result of the research and the experiments indicates that the method is reliable in theory, is feasible in technique. The precision of the optimal solution of problem can be enhanced with selecting proper threshold. The method is of high stability.
Lirong Yang - One of the best experts on this subject based on the ideXlab platform.
-
CenterMask: single shot instance segmentation with Point Representation
arXiv: Computer Vision and Pattern Recognition, 2020Co-Authors: Yuqing Wang, Hao Shen, Baoshan Cheng, Lirong YangAbstract:In this paper, we propose a single-shot instance segmentation method, which is simple, fast and accurate. There are two main challenges for one-stage instance segmentation: object instances differentiation and pixel-wise feature alignment. Accordingly, we decompose the instance segmentation into two parallel subtasks: Local Shape prediction that separates instances even in overlapping conditions, and Global Saliency generation that segments the whole image in a pixel-to-pixel manner. The outputs of the two branches are assembled to form the final instance masks. To realize that, the local shape information is adopted from the Representation of object center Points. Totally trained from scratch and without any bells and whistles, the proposed CenterMask achieves 34.5 mask AP with a speed of 12.3 fps, using a single-model with single-scale training/testing on the challenging COCO dataset. The accuracy is higher than all other one-stage instance segmentation methods except the 5 times slower TensorMask, which shows the effectiveness of CenterMask. Besides, our method can be easily embedded to other one-stage object detectors such as FCOS and performs well, showing the generalization of CenterMask.
-
CVPR - CenterMask: Single Shot Instance Segmentation With Point Representation
2020 IEEE CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020Co-Authors: Yuqing Wang, Hao Shen, Baoshan Cheng, Lirong YangAbstract:In this paper, we propose a single-shot instance segmentation method, which is simple, fast and accurate. There are two main challenges for one-stage instance segmentation: object instances differentiation and pixel-wise feature alignment. Accordingly, we decompose the instance segmentation into two parallel subtasks: Local Shape prediction that separates instances even in overlapping conditions, and Global Saliency generation that segments the whole image in a pixel-to-pixel manner. The outputs of the two branches are assembled to form the final instance masks. To realize that, the local shape information is adopted from the Representation of object center Points. Totally trained from scratch and without any bells and whistles, the proposed CenterMask achieves 34.5 mask AP with a speed of 12.3 fps, using a single-model with single-scale training/testing on the challenging COCO dataset. The accuracy is higher than all other one-stage instance segmentation methods except the 5 times slower TensorMask, which shows the effectiveness of CenterMask. Besides, our method can be easily embedded to other one-stage object detectors such as FCOS and performs well, showing the generation of CenterMask.
K. Sumita - One of the best experts on this subject based on the ideXlab platform.
-
A reinforcement learning scheme for acquisition of via-Point Representation of human motion
2004 IEEE International Joint Conference on Neural Networks (IEEE Cat. No.04CH37541), 1Co-Authors: Yasuhiro Wada, K. SumitaAbstract:Humans can generate a complex trajectory by imitative learning of others' movement. A method for learning complex sequential movements, but utilizing a via-Point Representation is proposed. However, the proposed algorithm for estimating a set of via-Points from the complex movement does not involve a learning process such as a learning by trial and error. The algorithm can find the minimum number of via-Points, and then specify the unique set of via-Points without a trial-and-error process. In this paper, we report an acquisition algorithm for via-Point Representation through trial and error in a human-like manner. The proposed via-Point acquisition algorithm based on reinforcement learning finds a set of via-Points that can mimic the reference trajectory by iterative learning using evaluation values of generated movement pattern.
Yuqing Wang - One of the best experts on this subject based on the ideXlab platform.
-
CenterMask: single shot instance segmentation with Point Representation
arXiv: Computer Vision and Pattern Recognition, 2020Co-Authors: Yuqing Wang, Hao Shen, Baoshan Cheng, Lirong YangAbstract:In this paper, we propose a single-shot instance segmentation method, which is simple, fast and accurate. There are two main challenges for one-stage instance segmentation: object instances differentiation and pixel-wise feature alignment. Accordingly, we decompose the instance segmentation into two parallel subtasks: Local Shape prediction that separates instances even in overlapping conditions, and Global Saliency generation that segments the whole image in a pixel-to-pixel manner. The outputs of the two branches are assembled to form the final instance masks. To realize that, the local shape information is adopted from the Representation of object center Points. Totally trained from scratch and without any bells and whistles, the proposed CenterMask achieves 34.5 mask AP with a speed of 12.3 fps, using a single-model with single-scale training/testing on the challenging COCO dataset. The accuracy is higher than all other one-stage instance segmentation methods except the 5 times slower TensorMask, which shows the effectiveness of CenterMask. Besides, our method can be easily embedded to other one-stage object detectors such as FCOS and performs well, showing the generalization of CenterMask.
-
CVPR - CenterMask: Single Shot Instance Segmentation With Point Representation
2020 IEEE CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020Co-Authors: Yuqing Wang, Hao Shen, Baoshan Cheng, Lirong YangAbstract:In this paper, we propose a single-shot instance segmentation method, which is simple, fast and accurate. There are two main challenges for one-stage instance segmentation: object instances differentiation and pixel-wise feature alignment. Accordingly, we decompose the instance segmentation into two parallel subtasks: Local Shape prediction that separates instances even in overlapping conditions, and Global Saliency generation that segments the whole image in a pixel-to-pixel manner. The outputs of the two branches are assembled to form the final instance masks. To realize that, the local shape information is adopted from the Representation of object center Points. Totally trained from scratch and without any bells and whistles, the proposed CenterMask achieves 34.5 mask AP with a speed of 12.3 fps, using a single-model with single-scale training/testing on the challenging COCO dataset. The accuracy is higher than all other one-stage instance segmentation methods except the 5 times slower TensorMask, which shows the effectiveness of CenterMask. Besides, our method can be easily embedded to other one-stage object detectors such as FCOS and performs well, showing the generation of CenterMask.
Long Long - One of the best experts on this subject based on the ideXlab platform.
-
Fixed-Point digital IIR filter design using two-stage ensemble evolutionary algorithm
Applied Soft Computing, 2013Co-Authors: Yu Wang, Thomas Weise, Long LongAbstract:The research on optimal design of infinite-impulse response (IIR) filter design based on various optimization techniques, including evolutionary algorithms (EAs), has gained much attention in recent years. Previously, the parameters of digital IIR filters are encoded with floating-Point Representations. It is known that a fixed-Point Representation can effectively save computational resources and is more convenient for direct realization on hardware. Inherently, compared with the floating-Point Representation, the fixed-Point Representation would make the search space miss much useful gradient information and therefore, surely rises new challenges for continuous EAs. In this paper, we first analyze the fitness landscape properties of optimal digital IIR filter design. Based on the fitness landscape investigation, a two-stage ensemble evolutionary algorithm (TEEA) is applied to digital IIR filter design with fixed-Point Representation. In order to fully evaluate the performance of TEEA, we experimentally compare it with five state-of-the-art EAs on four types of digital IIR filters with different settings. Based on the experimental results, we can conclude that TEEA has higher convergence speed, better exploration, and higher success rate. In order to benchmark TEEA further, we apply it to some more difficult problems with shorter word length or higher order. We can find that TEEA can provide satisfying performance on these hard tasks as well.