Using the first order and the second order derivative information of the network error function to the learning rate factor / / and the momentum factor ? , dynamic optimization of the learning rate factor / / and the momentum factor ? is obtained during the network training process , which efficiently speeds up the network learning rate 并利用網(wǎng)絡(luò)誤差函數(shù)對訓(xùn)練速率系數(shù)和沖量系數(shù)的一階和二階導(dǎo)數(shù)信息,在網(wǎng)絡(luò)訓(xùn)練過程中動(dòng)態(tài)優(yōu)化調(diào)整訓(xùn)練速率系數(shù)和沖量系數(shù),有效地加快了網(wǎng)絡(luò)的訓(xùn)練速度。
Aimed at the parameter estimation and model reduction problems of non - linear systems in noisy environment , a class of particle swami optimization ( pso ) approach with hypothesis test is proposed , named psoht , which estimates parameters by using pso operator in conjunction with evaluation and comparison in statistical sense to minimize mean square error function 摘要針對噪聲環(huán)境下的非線性系統(tǒng)參數(shù)估計(jì)和模型降階問題,提出了一種帶假設(shè)檢驗(yàn)的微粒群優(yōu)化算法( psoht ) ,以最小化平均平方誤差為目標(biāo),結(jié)合統(tǒng)計(jì)意義下的評(píng)價(jià)和比較,通過微粒群操作進(jìn)行參數(shù)估計(jì)。
The individual svm is prone to fail in the intrusion detection for the fragility of being attacked . this paper addresses a method using a support vector machines ensemble approach based on negative correlation learning for intrusion detection . using a correlation penalty term in the error function , the aggregate members can be accurate and diverse . and the evolutionary strategy is considered as the best way to automatically determine the individ . ual svms hyperparameters . at last we combine the results of all individual svms using ensemble technique . this distributed parallel detection can strengthen the robustness of the system . simulation results show the effectiveness of the method presented in this paper 在入侵檢測中使用單個(gè)的支持向量機(jī)容易因"單點(diǎn)失效"而危害系統(tǒng)安全.提出一種基于支持向量機(jī)集成的方法來進(jìn)行入侵檢測.它采用負(fù)相關(guān)學(xué)習(xí)技術(shù),在誤差項(xiàng)中使用相關(guān)性懲罰因子使得生成的分類器有更好的多樣性和精度;算法采用進(jìn)化策略來自動(dòng)地確定個(gè)體支持向量機(jī)的超參數(shù),避免了需要了解問題的先驗(yàn)知識(shí);最后,采用集成技術(shù)來組合個(gè)體支持向量機(jī)的檢測結(jié)果.仿真實(shí)驗(yàn)表明這一方法有更好的檢測性能,并且這種分布式并行檢測方法有利于增加入侵檢測系統(tǒng)的魯棒性
Two distinct bp training algorithms on multilayer perceptron type neural - network are developed to improve the trained network ' s quickness , robustness and generality . they are brought out from the combination of existing investigations - dynamic learning rate , momentum terms and quadratic function of weights , and amendment of desired error function respectively . the simulative results of rotating machinery fault pattern classification illustrated the effectiveness of the new algorithms 針對多層感知器神經(jīng)網(wǎng)絡(luò)的標(biāo)準(zhǔn)bp訓(xùn)練算法存在推廣能力差和訓(xùn)練速度慢等缺陷,首先結(jié)合已有的研究成果,給出了一種動(dòng)態(tài)學(xué)習(xí)率和動(dòng)量項(xiàng)相結(jié)合,并在誤差函數(shù)中引入網(wǎng)絡(luò)權(quán)值的二次型函數(shù)項(xiàng)的改進(jìn)訓(xùn)練算法,提高了網(wǎng)絡(luò)的快速性和推廣能力。
The load of the dc motor has a great influence on the tracking effect of speed control system , so a kind of load disturbance observer is constructed to compensate the load disturbance , and a new variable structure controller is designed by employing lyapunov speed error function 針對實(shí)際直流電機(jī)負(fù)載對速度控制系統(tǒng)跟蹤效果有較大影響的情況,提出了一種重構(gòu)軋制負(fù)載擾動(dòng)的觀測器,進(jìn)而對負(fù)載擾動(dòng)進(jìn)行前饋補(bǔ)償控制,設(shè)計(jì)了一種新的利用lyapunov速度誤差函數(shù)構(gòu)造的變結(jié)構(gòu)魯棒控制器。
In the design of corpus , we carefully analyze the syllable distribution of corpus th - coss , then classify the prosodic characters of this corpus and present out the distribution of every prosodic character . based on prosodic character vector , we construct an error function which is used to select original corpus for simulation system , and show the distribution of prosodic characters for the original corpus . greedy algorithm and corpus self - adaptive process are expatiated to set theoretical foundation for text material search 在語料庫分析與設(shè)計(jì)方面,首先統(tǒng)計(jì)th - coss語料庫中音節(jié)分布情況,給出th - coss語料庫韻律特征分類,并對每一種韻律特征進(jìn)行統(tǒng)計(jì),然后構(gòu)造了一個(gè)基于韻律特征向量的誤差函數(shù),并采用該誤差函數(shù)提取語料組成模擬系統(tǒng)的初始語料庫,分析該庫的韻律特征分布,最后闡述了greedy算法與語料自適應(yīng)過程,為文本語料的搜索打下理論基礎(chǔ)。
In order to improve the generalization ability and to make the system more accordant to the practical requirement , the paper introduces the k - l information distance to the error function of the traditional bp algorithm . the simulation proves that the generalization ability of the system trained with the modified algorithm is much better than that of other algorithms 為了進(jìn)一步改進(jìn)系統(tǒng)的泛化能力,使其滿足報(bào)價(jià)的實(shí)際需要,本文利用神經(jīng)網(wǎng)絡(luò)的概率描述,通過研究k - l信息距離和神經(jīng)網(wǎng)絡(luò)泛化能力的關(guān)系,構(gòu)造了一個(gè)新的神經(jīng)網(wǎng)絡(luò)學(xué)習(xí)誤差函數(shù),并將此法與其它各種算法的泛化結(jié)果進(jìn)行比較。