通用大语言模型(large language model,LLM)在数学古籍领域常因专业术语识别不准确、上下文关系理解偏差、知识推理不全和计算过程错误而产生严重的幻觉(hallucination)问题,包括事实性幻觉、忠实性幻觉和逻辑性幻觉。检索增强生成(retr...通用大语言模型(large language model,LLM)在数学古籍领域常因专业术语识别不准确、上下文关系理解偏差、知识推理不全和计算过程错误而产生严重的幻觉(hallucination)问题,包括事实性幻觉、忠实性幻觉和逻辑性幻觉。检索增强生成(retrieval-augmented generation,RAG)技术引入外部领域文本,目前是缓解大语言模型领域事实性幻觉的一种有效方法。然而RAG的检索内容存在噪声,且知识片段之间缺乏深层关联,使得RAG在缓解忠实性幻觉和逻辑性幻觉方面的能力较为有限。领域知识图谱(domain knowledge graph,DKG)具有将领域知识关联起来的能力,为此提出了DKG和RAG协同知识增强框架(collaborative knowledge augmentation between domain knowledge graph and retrieval-augmented generation,CogKAG),并构建数学古籍《九章算术》领域智能体。CogKAG框架智能体旨在通过检索DKG的结构化领域关联知识和利用RAG检索非结构化领域文本构建动态结构化综合上下文提示,增强LLM的上下文关系及逻辑推理计算能力,从而有效缓解其在数学古籍领域中的忠实性、逻辑性幻觉问题。实验结果证明,CogKAG框架智能体可以显著减轻LLM在数学古籍领域上的幻觉,从而提升在问答(question and answer,QA)任务上的性能。展开更多
Recently, unmanned aerial vehicles （UAVs） acting as relay platforms have attracted considerable attention due to the advantages of extending coverage and improving connectivity for long-range communications. Specifi...Recently, unmanned aerial vehicles （UAVs） acting as relay platforms have attracted considerable attention due to the advantages of extending coverage and improving connectivity for long-range communications. Specifically, in the scenario where the access point （AP） is mobile, a UAV needs to find an efficient path to guarantee the connectivity of the relay link. Motivated by this fact, this paper proposes an optimal design for beamforming （BF） and UAV path planning. First of all, we study a dual-hop amplify-and-forward （AF） wireless relay network, in which a UAV is used as relay between a mobile AP and a fixed base station （BS）. In the network, both of the AP and the BS are equipped with multiple antennas, whereas the UAV has a single antenna. Then, we obtain the output signal^to-noise ratio （SNR） of the dual-hop relay network. Based on the criterion of maximizing the output SNR, we develop an optimal design to obtain the solution of the optimal BF weight vector and the UAV heading angle. Next, we derive the closed-form outage probability （OP） expression to investigate the performance of the dual-hop relay network conveniently. Finally, computer simulations show that the proposed approach can obtain nearly optimal flying path and OP performance, indicating the effectiveness of the proposed algorithm. Furthermore, we find that increasing the antenna number at the BS or the maximal heading angle can significantly improve the performance of the considered relay network.展开更多
A multi-objective optimization based robust beamforming(BF)scheme is proposed to realize secure transmission in a cognitive satellite and unmanned aerial vehicle(UAV)network.Since the satellite network coexists with t...A multi-objective optimization based robust beamforming(BF)scheme is proposed to realize secure transmission in a cognitive satellite and unmanned aerial vehicle(UAV)network.Since the satellite network coexists with the UAV network,we first consider both achievable secrecy rate maximization and total transmit power minimization,and formulate a multi-objective optimization problem(MOOP)using the weighted Tchebycheff approach.Then,by supposing that only imperfect channel state information based on the angular information is available,we propose a method combining angular discretization with Taylor approximation to transform the non-convex objective function and constraints to the convex ones.Next,we adopt semi-definite programming together with randomization technology to solve the original MOOP and obtain the BF weight vector.Finally,simulation results illustrate that the Pareto optimal trade-off can be achieved,and the superiority of our proposed scheme is confirmed by comparing with the existing BF schemes.展开更多
Two clean liquid–phase cyclohexanone oxidation routes catalyzed by DHBEA and HTS zeolites, in the absence of organic solvents, have been developed for producing high value-added chemical intermediates. Under optimize...Two clean liquid–phase cyclohexanone oxidation routes catalyzed by DHBEA and HTS zeolites, in the absence of organic solvents, have been developed for producing high value-added chemical intermediates. Under optimized conditions,the cyclohexanone conversion reaches up to 60%, and the selectivity of total target products(ε-caprolactone, 6-hydroxyhexanoic acid and adipic acid) is over 90% achieved by the HTS zeolite; while both cyclohexanone conversion and the 6-hydroxyhexanoic acid selectivity are over 95% obtained on the DHBEA zeolite. Both the Lewis and Br鰊sted acid sites of DHBEA zeolite can preferentially activate the carbonyl group of cyclohexanone without any impact on H_2O_2 molecules.Meanwhile, the HTS zeolite can predominantly make H_2O_2 more reactive, which agrees well with the molecular calculation results. Hence, two different Baeyer-Villiger oxidation mechanisms based on the activation of H_2O_2 and cyclohexanone are proposed. Then, 6-hydroxyhexanoic acid is formed via the ring-opening of ε-caprolactone. However, C-OH groups cannot be reactivated by DHBEA zeolite, leading to insignificant adipic acid formation, while the selectivity of adipic acid is 28.5% obtained on the HTS zeolite. Consequently, the higher catalytic performance of the DHBEA zeolite is ascribed to its larger amount of active sites and greater diffusion features than those of HTS zeolite.展开更多
The toluene oxidative bromination reaction catalyzed by hollow titanium silicalite(HTS)zeolite in aqueous medium was investigated by employing H2O2 and HBr under mild conditions without the need for organic solvent.A ...The toluene oxidative bromination reaction catalyzed by hollow titanium silicalite(HTS)zeolite in aqueous medium was investigated by employing H2O2 and HBr under mild conditions without the need for organic solvent.A high toluene conversion(90.7%)and high selectivity of mono-bromotoluene(99.0%)was achieved under the optimal reaction conditions.The UV-Raman spectroscopy was applied for the mechanism study and the result reveals that HTS is efficient for catalyzing the oxidation reaction of HBr with H2O2 to produce abundant active bromine species,which can further facilitate the toluene electrophilic bromination reaction.A two-step toluene bromination reaction mechanism involving the HTS catalyzed active bromine species“generation-conversion-utilization”process is proposed based on the UV-Raman spectroscopy analysis.展开更多
文摘通用大语言模型(large language model,LLM)在数学古籍领域常因专业术语识别不准确、上下文关系理解偏差、知识推理不全和计算过程错误而产生严重的幻觉(hallucination)问题,包括事实性幻觉、忠实性幻觉和逻辑性幻觉。检索增强生成(retrieval-augmented generation,RAG)技术引入外部领域文本,目前是缓解大语言模型领域事实性幻觉的一种有效方法。然而RAG的检索内容存在噪声,且知识片段之间缺乏深层关联,使得RAG在缓解忠实性幻觉和逻辑性幻觉方面的能力较为有限。领域知识图谱(domain knowledge graph,DKG)具有将领域知识关联起来的能力,为此提出了DKG和RAG协同知识增强框架(collaborative knowledge augmentation between domain knowledge graph and retrieval-augmented generation,CogKAG),并构建数学古籍《九章算术》领域智能体。CogKAG框架智能体旨在通过检索DKG的结构化领域关联知识和利用RAG检索非结构化领域文本构建动态结构化综合上下文提示,增强LLM的上下文关系及逻辑推理计算能力,从而有效缓解其在数学古籍领域中的忠实性、逻辑性幻觉问题。实验结果证明,CogKAG框架智能体可以显著减轻LLM在数学古籍领域上的幻觉,从而提升在问答(question and answer,QA)任务上的性能。
基金supported by the National Natural Science Foundation of China (Nos. 61202351, 61271255)the Natural Science Foundation of Jiangsu Province (No. BK20131068)+2 种基金the Open Research Fund of National Mobile Communications Research Laboratory in Southeast University (No. 2012D15)the Funding of Jiangsu Innovation Program for Graduate Education (No. CXLX11_0202)the Fundamental Research Funds for the Central Universities
文摘Recently, unmanned aerial vehicles （UAVs） acting as relay platforms have attracted considerable attention due to the advantages of extending coverage and improving connectivity for long-range communications. Specifically, in the scenario where the access point （AP） is mobile, a UAV needs to find an efficient path to guarantee the connectivity of the relay link. Motivated by this fact, this paper proposes an optimal design for beamforming （BF） and UAV path planning. First of all, we study a dual-hop amplify-and-forward （AF） wireless relay network, in which a UAV is used as relay between a mobile AP and a fixed base station （BS）. In the network, both of the AP and the BS are equipped with multiple antennas, whereas the UAV has a single antenna. Then, we obtain the output signal^to-noise ratio （SNR） of the dual-hop relay network. Based on the criterion of maximizing the output SNR, we develop an optimal design to obtain the solution of the optimal BF weight vector and the UAV heading angle. Next, we derive the closed-form outage probability （OP） expression to investigate the performance of the dual-hop relay network conveniently. Finally, computer simulations show that the proposed approach can obtain nearly optimal flying path and OP performance, indicating the effectiveness of the proposed algorithm. Furthermore, we find that increasing the antenna number at the BS or the maximal heading angle can significantly improve the performance of the considered relay network.
基金supported by the Key International Cooperation Research Project(61720106003)the National Natural Science Foundation of China(62001517)+2 种基金the Shanghai Aerospace Science and Technology Innovation Foundation(SAST2019-095)the NUPTSF(NY220111)the Foundational Research Project of Complex Electronic System Simulation Laboratory(DXZT-JC-ZZ-2019-009,DXZTJC-ZZ-2019-005).
文摘A multi-objective optimization based robust beamforming(BF)scheme is proposed to realize secure transmission in a cognitive satellite and unmanned aerial vehicle(UAV)network.Since the satellite network coexists with the UAV network,we first consider both achievable secrecy rate maximization and total transmit power minimization,and formulate a multi-objective optimization problem(MOOP)using the weighted Tchebycheff approach.Then,by supposing that only imperfect channel state information based on the angular information is available,we propose a method combining angular discretization with Taylor approximation to transform the non-convex objective function and constraints to the convex ones.Next,we adopt semi-definite programming together with randomization technology to solve the original MOOP and obtain the BF weight vector.Finally,simulation results illustrate that the Pareto optimal trade-off can be achieved,and the superiority of our proposed scheme is confirmed by comparing with the existing BF schemes.
基金financially supported by the National Basic Research Program of China (973 Program, 2006CB202508)the Research Program of China Petrochemical Corporation (SINOPEC Group 20673054)the National Key Research and Development Program of China (2017YFB0306800)
文摘Two clean liquid–phase cyclohexanone oxidation routes catalyzed by DHBEA and HTS zeolites, in the absence of organic solvents, have been developed for producing high value-added chemical intermediates. Under optimized conditions,the cyclohexanone conversion reaches up to 60%, and the selectivity of total target products(ε-caprolactone, 6-hydroxyhexanoic acid and adipic acid) is over 90% achieved by the HTS zeolite; while both cyclohexanone conversion and the 6-hydroxyhexanoic acid selectivity are over 95% obtained on the DHBEA zeolite. Both the Lewis and Br鰊sted acid sites of DHBEA zeolite can preferentially activate the carbonyl group of cyclohexanone without any impact on H_2O_2 molecules.Meanwhile, the HTS zeolite can predominantly make H_2O_2 more reactive, which agrees well with the molecular calculation results. Hence, two different Baeyer-Villiger oxidation mechanisms based on the activation of H_2O_2 and cyclohexanone are proposed. Then, 6-hydroxyhexanoic acid is formed via the ring-opening of ε-caprolactone. However, C-OH groups cannot be reactivated by DHBEA zeolite, leading to insignificant adipic acid formation, while the selectivity of adipic acid is 28.5% obtained on the HTS zeolite. Consequently, the higher catalytic performance of the DHBEA zeolite is ascribed to its larger amount of active sites and greater diffusion features than those of HTS zeolite.
基金The author thanks for the financial support of SINOPEC Corporation(S413108).
文摘The toluene oxidative bromination reaction catalyzed by hollow titanium silicalite(HTS)zeolite in aqueous medium was investigated by employing H2O2 and HBr under mild conditions without the need for organic solvent.A high toluene conversion(90.7%)and high selectivity of mono-bromotoluene(99.0%)was achieved under the optimal reaction conditions.The UV-Raman spectroscopy was applied for the mechanism study and the result reveals that HTS is efficient for catalyzing the oxidation reaction of HBr with H2O2 to produce abundant active bromine species,which can further facilitate the toluene electrophilic bromination reaction.A two-step toluene bromination reaction mechanism involving the HTS catalyzed active bromine species“generation-conversion-utilization”process is proposed based on the UV-Raman spectroscopy analysis.
