针对油气领域知识图谱构建过程中命名实体识别使用传统方法存在实体特征信息提取不准确、识别效率低的问题,提出了一种基于BERT-BiLSTM-CRF模型的命名实体识别研究方法。该方法首先利用BERT(bidirectional encoder representations from...针对油气领域知识图谱构建过程中命名实体识别使用传统方法存在实体特征信息提取不准确、识别效率低的问题,提出了一种基于BERT-BiLSTM-CRF模型的命名实体识别研究方法。该方法首先利用BERT(bidirectional encoder representations from transformers)预训练模型得到输入序列语义的词向量;然后将训练后的词向量输入双向长短期记忆网络(bi-directional long short-term memory,BiLSTM)模型进一步获取上下文特征;最后根据条件随机场(conditional random fields,CRF)的标注规则和序列解码能力输出最大概率序列标注结果,构建油气领域命名实体识别模型框架。将BERT-BiLSTM-CRF模型与其他2种命名实体识别模型(BiLSTM-CRF、BiLSTM-Attention-CRF)在包括3万多条文本语料数据、4类实体的自建数据集上进行了对比实验。实验结果表明,BERT-BiLSTM-CRF模型的准确率(P)、召回率(R)和F_(1)值分别达到91.3%、94.5%和92.9%,实体识别效果优于其他2种模型。展开更多
Optical cryptanalysis is essential to the further investigation of more secure optical cryptosystems.Learning-based at-tack of optical encryption eliminates the need for the retrieval of random phase keys of optical e...Optical cryptanalysis is essential to the further investigation of more secure optical cryptosystems.Learning-based at-tack of optical encryption eliminates the need for the retrieval of random phase keys of optical encryption systems but it is limited for practical applications since it requires a large set of plaintext-ciphertext pairs for the cryptosystem to be at-tacked.Here,we propose a two-step deep learning strategy for ciphertext-only attack(COA)on the classical double ran-dom phase encryption(DRPE).Specifically,we construct a virtual DRPE system to gather the training data.Besides,we divide the inverse problem in COA into two more specific inverse problems and employ two deep neural networks(DNNs)to respectively learn the removal of speckle noise in the autocorrelation domain and the de-correlation operation to retrieve the plaintext image.With these two trained DNNs at hand,we show that the plaintext can be predicted in real-time from an unknown ciphertext alone.The proposed learning-based COA method dispenses with not only the retrieval of random phase keys but also the invasive data acquisition of plaintext-ciphertext pairs in the DPRE system.Numerical simulations and optical experiments demonstrate the feasibility and effectiveness of the proposed learning-based COA method.展开更多
In recent years,metasurfaces that enable the flexible wavefront modulation at sub-wavelength scale have been widely used into holographic display,due to its prominent advantages in polarization degrees of freedom,view...In recent years,metasurfaces that enable the flexible wavefront modulation at sub-wavelength scale have been widely used into holographic display,due to its prominent advantages in polarization degrees of freedom,viewing angle,and achromaticity in comparison with traditional holographic devices.In holography,the computational complexity of hologram,imaging sharpness,energy utilization,reproduction rate,and system indirection are all determined by the encoding method.Here,we propose a visible frequency broadband dielectric metahologram based on the random Fourier phase-only encoding method.Using this simple and convenient method,we design and fabricate a transmission-type geometric phase all-dielectric metahologram,which can realize holographic display with high quality in the visible frequency range.This method encodes the amplitude information into the phase function only once,eliminating the cumbersome iterations,which greatly simplifies the calculation process,and may facilitate the preparation of large area nanoprint-holograms.展开更多
文摘针对油气领域知识图谱构建过程中命名实体识别使用传统方法存在实体特征信息提取不准确、识别效率低的问题,提出了一种基于BERT-BiLSTM-CRF模型的命名实体识别研究方法。该方法首先利用BERT(bidirectional encoder representations from transformers)预训练模型得到输入序列语义的词向量;然后将训练后的词向量输入双向长短期记忆网络(bi-directional long short-term memory,BiLSTM)模型进一步获取上下文特征;最后根据条件随机场(conditional random fields,CRF)的标注规则和序列解码能力输出最大概率序列标注结果,构建油气领域命名实体识别模型框架。将BERT-BiLSTM-CRF模型与其他2种命名实体识别模型(BiLSTM-CRF、BiLSTM-Attention-CRF)在包括3万多条文本语料数据、4类实体的自建数据集上进行了对比实验。实验结果表明,BERT-BiLSTM-CRF模型的准确率(P)、召回率(R)和F_(1)值分别达到91.3%、94.5%和92.9%,实体识别效果优于其他2种模型。
基金financial supports from the National Natural Science Foundation of China(NSFC)(62061136005,61705141,61805152,61875129,61701321)Sino-German Research Collaboration Group(GZ 1391)+2 种基金the Mobility program(M-0044)sponsored by the Sino-German CenterChinese Academy of Sciences(QYZDB-SSW-JSC002)Science and Technology Innovation Commission of Shenzhen(JCYJ20170817095047279)。
文摘Optical cryptanalysis is essential to the further investigation of more secure optical cryptosystems.Learning-based at-tack of optical encryption eliminates the need for the retrieval of random phase keys of optical encryption systems but it is limited for practical applications since it requires a large set of plaintext-ciphertext pairs for the cryptosystem to be at-tacked.Here,we propose a two-step deep learning strategy for ciphertext-only attack(COA)on the classical double ran-dom phase encryption(DRPE).Specifically,we construct a virtual DRPE system to gather the training data.Besides,we divide the inverse problem in COA into two more specific inverse problems and employ two deep neural networks(DNNs)to respectively learn the removal of speckle noise in the autocorrelation domain and the de-correlation operation to retrieve the plaintext image.With these two trained DNNs at hand,we show that the plaintext can be predicted in real-time from an unknown ciphertext alone.The proposed learning-based COA method dispenses with not only the retrieval of random phase keys but also the invasive data acquisition of plaintext-ciphertext pairs in the DPRE system.Numerical simulations and optical experiments demonstrate the feasibility and effectiveness of the proposed learning-based COA method.
基金supported by the National Natural Science Foundation of China(Grant Nos.11634010,91850118,11774289,61675168,and 11804277)the National Key Research and Development Program of China(Grant No.2017YFA0303800)the Fundamental Research Funds for the Central Universities(Grant Nos.3102018zy036,3102019JC008,and 310201911cx022)。
文摘In recent years,metasurfaces that enable the flexible wavefront modulation at sub-wavelength scale have been widely used into holographic display,due to its prominent advantages in polarization degrees of freedom,viewing angle,and achromaticity in comparison with traditional holographic devices.In holography,the computational complexity of hologram,imaging sharpness,energy utilization,reproduction rate,and system indirection are all determined by the encoding method.Here,we propose a visible frequency broadband dielectric metahologram based on the random Fourier phase-only encoding method.Using this simple and convenient method,we design and fabricate a transmission-type geometric phase all-dielectric metahologram,which can realize holographic display with high quality in the visible frequency range.This method encodes the amplitude information into the phase function only once,eliminating the cumbersome iterations,which greatly simplifies the calculation process,and may facilitate the preparation of large area nanoprint-holograms.
