<button id="umcoe"></button> A good review of the k-meanS algorithm
標簽： algorithm k-meanS review good
上傳時間： 2014-01-11
上傳用戶：xinyuzhiqiwuwu
This is the code for vector quantization by using k-meanS
標簽： quantization k-meanS vector using
上傳時間： 2017-08-22
上傳用戶：ynsnjs
Mean shift clustering. K meanS clustering.
標簽： clustering shift meanS Mean
上傳時間： 2014-01-08
上傳用戶：釣鰲牧馬
My version of k-meanS function. Improved so that there are no empty clusters after segmentation.
標簽： segmentation Improved function clusters
上傳時間： 2013-12-24
上傳用戶：hoperingcong
介紹K-meanS類型算法的基本原理，發(fā)展以及趨勢。
上傳時間： 2014-01-02
上傳用戶：1583060504
Hadoop下的K-meanS的Java實現(xiàn)，使用eclipse。
上傳時間： 2017-05-12
上傳用戶：18335103184
K-meanS算法是最古老也是應用最廣泛的聚類算法，它使用質(zhì)心定義原型，質(zhì)心是一組點的均值，通常該算法用于n維連續(xù)空間中的對象。 K-meanS算法流程 step1：選擇K個點作為初始質(zhì)心 step2：repeat 將每個點指派到最近的質(zhì)心，形成K個簇 重新計算每個簇的質(zhì)心 until 質(zhì)心不在變化 例如下圖的樣本集，初始選擇是三個質(zhì)心比較集中，但是迭代3次之后，質(zhì)心趨于穩(wěn)定，并將樣本集分為3部分 我們對每一個步驟都進行分析 step1：選擇K個點作為初始質(zhì)心 這一步首先要知道K的值，也就是說K是手動設(shè)置的，而不是像EM算法那樣自動聚類成n個簇 其次，如何選擇初始質(zhì)心 最簡單的方式無異于，隨機選取質(zhì)心了，然后多次運行，取效果最好的那個結(jié)果。這個方法，簡單但不見得有效，有很大的可能是得到局部最優(yōu)。 另一種復雜的方式是，隨機選取一個質(zhì)心，然后計算離這個質(zhì)心最遠的樣本點，對于每個后繼質(zhì)心都選取已經(jīng)選取過的質(zhì)心的最遠點。使用這種方式，可以確保質(zhì)心是隨機的，并且是散開的。 step2：repeat 將每個點指派到最近的質(zhì)心，形成K個簇 重新計算每個簇的質(zhì)心 until 質(zhì)心不在變化 如何定義最近的概念，對于歐式空間中的點，可以使用歐式空間，對于文檔可以用余弦相似性等等。對于給定的數(shù)據(jù)，可能適應與多種合適的鄰近性度量。
標簽： K-meanS Java 數(shù)據(jù)挖掘 聚類 算法
上傳時間： 2018-11-27
上傳用戶：1159474180
Methods for designing a maintenance simulation training system for certain kind of radio are introduced. Fault modeling method is used to establish the fault database. The system sets up some typical failures， follow the prompts trainers can locate the fault source and confirm the type to accomplish corresponding fault maintenance training. A training evaluation meanS is given to examining and evaluating the training performance. The system intuitively and vividly shows the fault maintenance process， it can not only be used in teaching， but also in daily maintenance training to efficiently improve the maintenance operation level. Graphical programming language LabVIEW is used to develop the system platform.
標簽： 電臺維修 模擬訓練 方法研究 系統(tǒng)設(shè)計
上傳時間： 2013-11-19
上傳用戶：3294322651
Providing power for the Pentium® microprocessor family isnot a trivial task by any meanS. In an effort to simplify thistask we have developed a new switching regulator controlcircuit and a new linear regulator to address the needs ofthese processors. Considerable time has been spent developingan optimized decoupling network. Here are severalcircuits using the new LTC®1266 synchronous buck regulatorcontrol chip and the LT®1584 linear regulator toprovide power for Pentium processors and Pentium VREprocessors. The Pentium processor has a supply requirementof 3.3V ±5%. The Pentium VRE processor requires3.500V ±100mV.
上傳時間： 2013-11-01
上傳用戶：名爵少年
鎖定放大是微弱信號檢測的重要手段。基于相關(guān)檢測理論，利用開關(guān)電容的開關(guān)實現(xiàn)鎖定放大器中乘法器的功能，提出開關(guān)電容和積分器相結(jié)合以實現(xiàn)相關(guān)檢測的方法，并設(shè)計出一種鎖定放大器。該鎖定放大器將微弱信號轉(zhuǎn)化為與之相關(guān)的方波，通過后續(xù)電路得到正比于被測信號的直流電平，為后續(xù)采集處理提供方便。測量數(shù)據(jù)表明鎖定放大器前級可將10-6 A的電流轉(zhuǎn)換為10-1 V的電壓，后級通過帶通濾波器級聯(lián)可將信號放大1×105倍。該方法在降低噪聲的同時，可對微弱信號進行放大，線性度較高、穩(wěn)定性較好。 Abstract: Lock-in Amplifying（LIA）is one of important meanS for weak signal detection. Based on cross-correlation detection theory， switch in the swithched capacitor was used as multiplier of LIA， and a new method of correlation detection was proposed combining swithched capacitor with integrator. A kind of LIA was designed which can convert the weak signal to square-wave， then DC proportional to measured signal was obtained through follow-up conditioning circuit， providing convenience for signal acquisition and processing. The measured data shows that the electric current（10-6 A） can be changed into voltage（10-1 V） by LIA， and the signal is magnified 1×105 times by cascade band-pass filter. The noise is suppressed and the weak signal is amplified. It has the advantages of good linearity and stability.
標簽： 開關(guān)電容 鎖定放大器
上傳時間： 2013-11-29
上傳用戶：黑漆漆
蟲蟲下載站版權(quán)所有 京ICP備2021023401號-1
