indoornewton

可用於arima的計算,檢測數據以及預測用途(do arima)

This code will find the best neural network by training a lot of times and calculating the MSE

最小均方误差（MMSE）的算法，从先验概率中采样，计算权重，包括调制，解调，信噪比计算。( Minimum mean square error (MMSE) algorithm, Sampling a priori probability, calculate the weight, Includes the modulation, demodulation, signal to noise ratio calculation.)

本程序介绍了相关算法在干涉仪侧向上的应用，具有很好的实时性。(This procedure describes the lateral correlation algorithm in the application of interferometer has good real-time.)

说明：  对给定概率或已知概率的信息源进行哈弗曼编码(Probability of a given or known probability of information sources Havermann coding)

car tracking with kalman filterring

关于麦克风阵列信号处理，根据声波传达理论，利用多个麦克风收集的信号可以将某一方向传来的声音增强或抑制，通过改变权值来改变某一方向的信号的接收强弱(About microphone array signal processing , according to the theory of sound waves to communicate using signals multiple microphones to collect a certain direction came the sound can be enhanced or inhibited by changing the weights to change the receive signal strength in a certain direction )

Minhashing和LSH算法。用于查找和比较一个pair是否是相似的。并分析false negative和false positive的值(Minhashing and Locality-Sensitive-Hashing (LSH). The algorithms are approximate in that they find only candidate pairs that are likely similar. Therefore, there are two types of error. A false positive (FP) is a candidate pair that is not actually similar according to the given similarity threshold. A false negative (FN) is a similar pair that is not a candidate pair. In this project, we will implement the Minhashing and LSH algorithms, apply them on data sets, and draw observations about when these algorithms perform. )

introduction to pattern recognition

commonly used algorithm