Principal Component Analysis – PCA is the process of projecting variance of data using minimal dimensions. The direction of dimension along which the variance is maximised(highest) will be chosen as the first principal component. The next best direction as the second P.C. and should be orthogonal to the first P.C,Continue Reading
Auto Encoders try to find the relations of the input with itself. These relations able the Auto Encoder algorithm to reproduce the same input via the relations it found. Auto Encoders – try to find the identity function f(x) using the latent factors of the data via f(x) = g(L)Continue Reading
Over Fitting in ML – occurs often because neural networks tend to be high-variance. Some techniques can minimize this variance by: 1. Regularization: Regularization adds a penalty on the different parameters of the model to reduce the freedom of the model. Hence, the model will be less likely to fitContinue Reading
k-means clustering is a method of vector quantization which finds good inter similarity between cluster members and intra-similarity between other clusters. k-means clustering partitions n observations into k clusters in which each observation belongs to the cluster with the nearest mean, These means are chosen so that in the nextContinue Reading
The cross-entropy measure is the sum of all the natural logs of each data point’s probability in accordance to the given or calculated labels divided by the number of points times (-1){so it can be positive number}. It’s simple formula given N points: (When y(n) is 1 the first term ofContinue Reading
The softmax function A.K.A normalized exponential function, squashes a K-dimensional input vector into a K-dimensional output vector of real values where each entry is in the range (0, 1). It’s like a sigmoid function, but it also divides each output such that the total sum of the outputs add upContinue Reading
The ROC curve(relative operating characteristic curve), is a comparison of two operating characteristics (TPR and FPR) as the criterion changes (sensitivity as a function of fall-out). The TPR(true positive rate) is also known as sensitivity, recall or probability of detection, it can be calculated as 1-miss rate(FNR). The FPR(false positiveContinue Reading
In image processing, a kernel, convolution matrix, or mask is a small matrix (usually 3X3 but can be any matrix size). It is used for blurring, sharpening, embossing, edge detection, and more. This is accomplished by doing a convolution between a kernel and an image. Convolution is termed as theContinue Reading
Non-maximum suppression is used in CV in object detection where multiple probabilities that are above the given threshold are detected(classification result). Thus non-maximum suppression can help to suppress all the gradient values (by setting them to 0) except the local maxima classification result. If the strength of the current pixel/elemntContinue Reading
Pooling is a form of non-linear down-sampling(image reduction) with an aggregation function. Max pooling is the most common – It partitions the input image into a set of non-overlapping rectangles(N x N) and, for each such sub-region, outputs the maximum, thus, practically extracting the most significant features. The non-overlapping poolingContinue Reading