Our previous work has introduced the newly generated program using the code transformation model GPT-2,verifying the generated programming codes through simhash(SH)and longest common subsequence(LCS)algo-rithms.Howeve...Our previous work has introduced the newly generated program using the code transformation model GPT-2,verifying the generated programming codes through simhash(SH)and longest common subsequence(LCS)algo-rithms.However,the entire code transformation process has encountered a time-consuming problem.Therefore,the objective of this study is to speed up the code transformation process signicantly.This paper has proposed deep learning approaches for modifying SH using a variational simhash(VSH)algorithm and replacing LCS with a piecewise longest common subsequence(PLCS)algorithm to faster the verication process in the test phase.Besides the code transformation model GPT-2,this study has also introduced MicrosoMASS and Facebook BART for a comparative analysis of their performance.Meanwhile,the explainable AI technique using local interpretable model-agnostic explanations(LIME)can also interpret the decision-making ofAImodels.The experimental results show that VSH can reduce the number of qualied programs by 22.11%,and PLCS can reduce the execution time of selected pocket programs by 32.39%.As a result,the proposed approaches can signicantly speed up the entire code transformation process by 1.38 times on average compared with our previous work.展开更多
建设智能教育平台是推动教育智能化的一个重要过程,但智能教育平台依赖的人工智能模型在训练过程中会消耗大量电力,因此,开展短期电力负荷预测对建设智能教育平台具有重要意义.针对在考虑多个属性开展短期电力负荷预测时,由于部分属性...建设智能教育平台是推动教育智能化的一个重要过程,但智能教育平台依赖的人工智能模型在训练过程中会消耗大量电力,因此,开展短期电力负荷预测对建设智能教育平台具有重要意义.针对在考虑多个属性开展短期电力负荷预测时,由于部分属性与电力负荷数据的相关性不强并且Transformer无法捕捉电力负荷数据的时间相关性,而导致电力负荷预测不够准确的问题,基于SR(Székely and Rizzo)距离相关系数、融合时间定位编码和Transformer,提出了一种短期电力负荷预测模型SF-Transformer.SF-Transformer通过SR距离相关系数对影响电力负荷数据的属性进行筛选,选择与电力负荷数据之间SR距离相关系数较大的属性.SF-Transformer采用一种全局时间编码与局部位置编码相结合的融合时间定位编码,有助于模型全面获取电力负荷数据的时间定位信息.在数据集上开展了实验,实验结果表明SF-Transformer与其他模型相比,在两种时长上进行电力负荷预测具有更低的均方根误差和平均绝对误差.展开更多
When source code is over-specific to some concrete contexts, developers have to manually change the source code retrieved from the Internet. To solve this problem, we propose the context-aware change pattern(CACP). ...When source code is over-specific to some concrete contexts, developers have to manually change the source code retrieved from the Internet. To solve this problem, we propose the context-aware change pattern(CACP). For a piece of source code, we extract the changes and changes-relevant context from the past code changes, identifying CACP that is the abstract common part of the changes and context. By using CACP, the retrieved source code could be transformed into the suitable one according to different user needs. From the Github we extracted 7 topics, collected 5-6 code snippets per topic and performed 5 different experiments which illustrated that CACP improves code transformation accuracy by 73.84%.展开更多
Walsh-Hadamard transform (WriT) can solve linear error equations on Field F2, and the method can be used to recover the parameters of convolutional code. However, solving the equations with many unknowns needs enorm...Walsh-Hadamard transform (WriT) can solve linear error equations on Field F2, and the method can be used to recover the parameters of convolutional code. However, solving the equations with many unknowns needs enormous computer memory which limits the application of WriT. In order to solve this problem, a method based on segmented WriT is proposed in this paper. The coefficient vector of high dimension is reshaped and two vectors of lower dimension are obtained. Then the WriT is operated and the requirement for computer memory is much reduced. The code rate and the constraint length of convolutional code are detected from the Walsh spectrum. And the check vector is recovered from the peak position. The validity of the method is verified by the simulation result, and the performance is proved to be optimal.展开更多
In order to solve the recognition of polyphase code radar signal, this paper gives two methods based on Frank code, i.e. the high-order spectrum recognition method and the fractional Fourier transform (FRFT) method, b...In order to solve the recognition of polyphase code radar signal, this paper gives two methods based on Frank code, i.e. the high-order spectrum recognition method and the fractional Fourier transform (FRFT) method, by analyzing the micro characteristics of polyphase code signals in time and frequency domain respectively. And a recognition algorithm based on Wigner-Hough transform (WHT) is developed in this paper. We verify the validity of each method by computer simulation and give relative merits and demerits. A set of results demonstrate that the algorithm based on Wigner-Hough transform has better recognition performance in low signal-to-noise (SNR) than others.展开更多
Luby transform (LT) codes are proposed to suppress the effect of partial band noise jam- ming in frequency hopping (FH) communication systems. A decoding scheme for joint erasures of severely jammed symbols and er...Luby transform (LT) codes are proposed to suppress the effect of partial band noise jam- ming in frequency hopping (FH) communication systems. A decoding scheme for joint erasures of severely jammed symbols and error correction is proposed. If an uncorrectable error is detected, the receiver erases the jammed symbols and uses incremental redundancy to increase the error-correcting capability. The performance of LT codes, under power-oppressive partial band noise jamming ( PB- N J) with the additive white Gaussian noise (AWGN), is evaluated via simulation. Even if the jam- mer spreads its high power over half of the hopping bandwidth, LT codes are shown to achieve a tar- get bit error probability of 10 -5, demonstrating their effectiveness as high-performance codes to im- prove the ability of FH systems to combat varying partial band noise jamming.展开更多
In order to improve the acquisition probability of satellite navigation signals, this paper proposes a novel code acquisition method based on wavelet transform filtering. Firstly, the signal vector based on the signal...In order to improve the acquisition probability of satellite navigation signals, this paper proposes a novel code acquisition method based on wavelet transform filtering. Firstly, the signal vector based on the signal passing through a set of partial matched filters (PMFs) is built. Then, wavelet domain filtering is performed on the signal vector value. Since the correlation signal is low in frequency and narrow in bandwidth, the noise out-of-band can be filtered out and the most of the useful signal energy is retained. Thus this process greatly improves the signal to noise ratio (SNR). Finally, the detection variable when the filtered signal goes through the combination process is constructed and the detection based on signal energy is made. Moreover, for the better retaining useful signal energy, the rule of selection of wavelet function has been made. Simulation results show the proposed method has a better detection performance than the normal code acquisition methods under the same false alarm probability.展开更多
为了解决Transformer编码器在行人重识别中因图像块信息丢失以及行人局部特征表达不充分导致模型识别准确率低的问题,本文提出改进型Transformer编码器和特征融合的行人重识别算法。针对Transformer在注意力运算时会丢失行人图像块相对...为了解决Transformer编码器在行人重识别中因图像块信息丢失以及行人局部特征表达不充分导致模型识别准确率低的问题,本文提出改进型Transformer编码器和特征融合的行人重识别算法。针对Transformer在注意力运算时会丢失行人图像块相对位置信息的问题,引入相对位置编码,促使网络关注行人图像块语义化的特征信息,以增强行人特征的提取能力。为了突出包含行人区域的显著特征,将局部patch注意力机制模块嵌入到Transformer网络中,对局部关键特征信息进行加权强化。最后,利用全局与局部信息特征融合实现特征间的优势互补,提高模型识别能力。训练阶段使用Softmax及三元组损失函数联合优化网络,本文算法在Market1501和DukeMTMC⁃reID两大主流数据集中评估测试,Rank⁃1指标分别达到97.5%和93.5%,平均精度均值(mean Average precision,mAP)分别达到92.3%和83.1%,实验结果表明改进型Transformer编码器和特征融合算法能够有效提高行人重识别的准确率。展开更多
基金supported by the Ministry of Science and Technology,Taiwan,under Grant Nos.MOST 111-2221-E-390-012 and MOST 111-2622-E-390-001.
文摘Our previous work has introduced the newly generated program using the code transformation model GPT-2,verifying the generated programming codes through simhash(SH)and longest common subsequence(LCS)algo-rithms.However,the entire code transformation process has encountered a time-consuming problem.Therefore,the objective of this study is to speed up the code transformation process signicantly.This paper has proposed deep learning approaches for modifying SH using a variational simhash(VSH)algorithm and replacing LCS with a piecewise longest common subsequence(PLCS)algorithm to faster the verication process in the test phase.Besides the code transformation model GPT-2,this study has also introduced MicrosoMASS and Facebook BART for a comparative analysis of their performance.Meanwhile,the explainable AI technique using local interpretable model-agnostic explanations(LIME)can also interpret the decision-making ofAImodels.The experimental results show that VSH can reduce the number of qualied programs by 22.11%,and PLCS can reduce the execution time of selected pocket programs by 32.39%.As a result,the proposed approaches can signicantly speed up the entire code transformation process by 1.38 times on average compared with our previous work.
文摘建设智能教育平台是推动教育智能化的一个重要过程,但智能教育平台依赖的人工智能模型在训练过程中会消耗大量电力,因此,开展短期电力负荷预测对建设智能教育平台具有重要意义.针对在考虑多个属性开展短期电力负荷预测时,由于部分属性与电力负荷数据的相关性不强并且Transformer无法捕捉电力负荷数据的时间相关性,而导致电力负荷预测不够准确的问题,基于SR(Székely and Rizzo)距离相关系数、融合时间定位编码和Transformer,提出了一种短期电力负荷预测模型SF-Transformer.SF-Transformer通过SR距离相关系数对影响电力负荷数据的属性进行筛选,选择与电力负荷数据之间SR距离相关系数较大的属性.SF-Transformer采用一种全局时间编码与局部位置编码相结合的融合时间定位编码,有助于模型全面获取电力负荷数据的时间定位信息.在数据集上开展了实验,实验结果表明SF-Transformer与其他模型相比,在两种时长上进行电力负荷预测具有更低的均方根误差和平均绝对误差.
基金Supported by the National Natural Science Foundation of China(60373075,61640221,61562026,61672470)
文摘When source code is over-specific to some concrete contexts, developers have to manually change the source code retrieved from the Internet. To solve this problem, we propose the context-aware change pattern(CACP). For a piece of source code, we extract the changes and changes-relevant context from the past code changes, identifying CACP that is the abstract common part of the changes and context. By using CACP, the retrieved source code could be transformed into the suitable one according to different user needs. From the Github we extracted 7 topics, collected 5-6 code snippets per topic and performed 5 different experiments which illustrated that CACP improves code transformation accuracy by 73.84%.
基金supported by the National Natural Science Foundation of China(61072120)
文摘Walsh-Hadamard transform (WriT) can solve linear error equations on Field F2, and the method can be used to recover the parameters of convolutional code. However, solving the equations with many unknowns needs enormous computer memory which limits the application of WriT. In order to solve this problem, a method based on segmented WriT is proposed in this paper. The coefficient vector of high dimension is reshaped and two vectors of lower dimension are obtained. Then the WriT is operated and the requirement for computer memory is much reduced. The code rate and the constraint length of convolutional code are detected from the Walsh spectrum. And the check vector is recovered from the peak position. The validity of the method is verified by the simulation result, and the performance is proved to be optimal.
文摘In order to solve the recognition of polyphase code radar signal, this paper gives two methods based on Frank code, i.e. the high-order spectrum recognition method and the fractional Fourier transform (FRFT) method, by analyzing the micro characteristics of polyphase code signals in time and frequency domain respectively. And a recognition algorithm based on Wigner-Hough transform (WHT) is developed in this paper. We verify the validity of each method by computer simulation and give relative merits and demerits. A set of results demonstrate that the algorithm based on Wigner-Hough transform has better recognition performance in low signal-to-noise (SNR) than others.
基金Supported by the National Nature Science Foundation of China(61072048)
文摘Luby transform (LT) codes are proposed to suppress the effect of partial band noise jam- ming in frequency hopping (FH) communication systems. A decoding scheme for joint erasures of severely jammed symbols and error correction is proposed. If an uncorrectable error is detected, the receiver erases the jammed symbols and uses incremental redundancy to increase the error-correcting capability. The performance of LT codes, under power-oppressive partial band noise jamming ( PB- N J) with the additive white Gaussian noise (AWGN), is evaluated via simulation. Even if the jam- mer spreads its high power over half of the hopping bandwidth, LT codes are shown to achieve a tar- get bit error probability of 10 -5, demonstrating their effectiveness as high-performance codes to im- prove the ability of FH systems to combat varying partial band noise jamming.
基金supported by the National Natural Science Foundation of China(6117213861401340)the Fundamental Research Funds for the Central Universities(K5051302015)
文摘In order to improve the acquisition probability of satellite navigation signals, this paper proposes a novel code acquisition method based on wavelet transform filtering. Firstly, the signal vector based on the signal passing through a set of partial matched filters (PMFs) is built. Then, wavelet domain filtering is performed on the signal vector value. Since the correlation signal is low in frequency and narrow in bandwidth, the noise out-of-band can be filtered out and the most of the useful signal energy is retained. Thus this process greatly improves the signal to noise ratio (SNR). Finally, the detection variable when the filtered signal goes through the combination process is constructed and the detection based on signal energy is made. Moreover, for the better retaining useful signal energy, the rule of selection of wavelet function has been made. Simulation results show the proposed method has a better detection performance than the normal code acquisition methods under the same false alarm probability.
文摘为了解决Transformer编码器在行人重识别中因图像块信息丢失以及行人局部特征表达不充分导致模型识别准确率低的问题,本文提出改进型Transformer编码器和特征融合的行人重识别算法。针对Transformer在注意力运算时会丢失行人图像块相对位置信息的问题,引入相对位置编码,促使网络关注行人图像块语义化的特征信息,以增强行人特征的提取能力。为了突出包含行人区域的显著特征,将局部patch注意力机制模块嵌入到Transformer网络中,对局部关键特征信息进行加权强化。最后,利用全局与局部信息特征融合实现特征间的优势互补,提高模型识别能力。训练阶段使用Softmax及三元组损失函数联合优化网络,本文算法在Market1501和DukeMTMC⁃reID两大主流数据集中评估测试,Rank⁃1指标分别达到97.5%和93.5%,平均精度均值(mean Average precision,mAP)分别达到92.3%和83.1%,实验结果表明改进型Transformer编码器和特征融合算法能够有效提高行人重识别的准确率。