machine learning

machine learning

Fully Connected Layer: The brute force layer of a Machine Learning model

Fully Connected layers in a neural networks are those layers where all the inputs from one layer are connected to every activation unit of the next layer. In most popular machine learning models, the last few layers are full connected layers which compiles the data extracted by previous layers

machine learning

Convolution Layer: The layer that takes over 70% of time in a Machine Learning model

Convolutional Layer is the most important layer in a Machine Learning model where the important features from the input are extracted and where most of the computational time (>=70% of the total inference time) is spent. Concepts involved are kernel size, padding, feature map and strides

clustering algorithm

K+ Means Clustering algorithm

K+ Means algorithm is a clustering algorithm and an improvement to K means clustering algorithm and solves the problem of choosing K (number of clusters). It is great at detecting outliers and forming new clusters. The complexity is O(t*(k^2)*n) which is slightly more than K means algorithm

machine learning

Types of Data Formats in Machine Learning

Each data format represents how the input data is represented in memory. This is important as each machine learning application performs well for a particular data format and worse for others. Various data formats are NHWC, NCHW, NCDHW and NDHWC

machine learning

Types of Loss Functions in Machine Learning

The various types of loss functions are mean_squared_error, mean_absolute_error, mean_absolute_percentage_error, mean_squared_logarithmic_error, squared_hinge, hinge, categorical_hinge, logcosh, categorical_crossentropy, sparse categorical / binary crossentropy, kullback_leibler_divergence and other

machine learning

Bayesian model

A Bayesian model is a statistical model where we use probability to represent both the uncertainty regarding the output and input to the model. The basic idea is that we start by assuming something which is adjusted based upon input data. We look into Bayesian Linear Regression as well

machine learning

Sentiment Analysis using LSTM with Keras

Sentimental analysis is one of the most important applications of Machine learning. It is used extensively in Netflix and YouTube to suggest videos, Google Search and others. In this article, we will build a sentiment analyser from scratch using KERAS framework with Python using concepts of LSTM.

machine learning

Long Short Term Memory (LSTM)

Long short-term memory (LSTM) units are units of a recurrent neural network (RNN). An RNN composed of LSTM units is often called an LSTM network. A common LSTM unit is composed of a cell, an input gate, an output gate and a forget gate. It has applications in Speech recognition, Video synthesis

machine learning

Hierarchical Clustering

Hierarchical clustering is a method of clustering. In this method, we find a hierarchy of clusters which looks like the hierarchy of folders in your operating system. This hierarchy of clusters will resemble a tree structure and it is called dendrogram

machine learning

Kernel Principal Component Analysis (KPCA)

Kernel Principal Component Analysis (KPCA) is a non-linear dimensionality reduction technique. It is an extension of Principal Component Analysis (PCA) - which is a linear dimensionality reduction technique - using kernel methods.

machine learning

Convolutional Neural Networks (CNN)

Convolutional Neural Network (CNN) is an neural network which extracts or identifies a feature in a particular image and is the basis of GoogleNet and VGG19 and used for object detection and classification. CNN has five basic components Convolution, ReLU, Pooling, Flattening and Full connection.

machine learning

Understand basic TensorFlow programming concepts

We understand basic TensorFlow programming concepts using a code to add two numbers in TensorFlow. We created placeholders like tf.placeholder, operations like tf.constant and tf.add, sessions using tf.session and configuration using tf.configProto

machine learning

Training vs Inference

Training and Inference are two major processes of Machine Learning and is deeply connected. Training is the process by which we generate various parameters such as weights and biases which are used in a particular model. Inference is the process of using the trained model to do a particular task

machine learning

Hidden Markov Model

Hidden Markov Model is a stochastic model describing a sequence of possible events in which the probability of each event depends on the state attained in the previous event. Markov model can be used in real life forecasting problems. Simple Markov model cannot be used for customer level predictions

machine learning

Markov Chain

Markov chain is a stochastic model describing a sequence of possible events in which the probability of each event depends only on the state attained in the previous event. It refers to the sequence of random variables such a process moves through, with the Markov property of serial dependence

machine learning

Deep Learning on 2-Dimensional Images

Applying deep learning concepts on images has proved to be one of the important work which has resulted in early detection of diseases resulting in saving millions of life to monitoring activities on the entire Earth We take a look at medical images, Satellite Images and the various Python libraries

machine learning

Deep Learning for Medical Imaging and Diagnosis

One of the major medical challenges that we face today is the early detection of diseases so that the proper threatment can be applied. This can be solved by applying machine learning to analyse MRI scan, CT Scan, Xray Scans, InfraRed Images, Arthoscopy, UV radiation, Scintigraphy and Ultrasound

machine learning

Recurrent Neural Networks (RNN)

Recurrent Neural Network is one of the widely used algorithms of Deep Learning mainly due to is unique Design. It is the only algorithm that remembers the most recent Input and makes use of memory element. It is used by Apple Siri and Google Voice Search. RNN is used for sequential data.

machine learning

Residual Network (ResNet)

ResNet makes it possible to train up to hundreds or even thousands of layers and still achieves compelling performance. Thanks to this technique they were able to train a network with 152 layers while still having lower complexity than VGGNet. It achieves a top-5 error rate of 3.57%

machine learning

GoogleNet / InceptionNet

The winner of the ILSVRC 2014 competition was GoogleNet from Google. It achieved a top-5 error rate of 6.67%. GoogleNet has 22 layer, and almost 12x less parameters (So faster and less then Alexnet and much more accurate). Their idea was to make a model that also could be used on a smart-phone

machine learning

Architecture of AlexNet and its current use

Alexnet is a Deep Convolutional Neural Network (CNN) for image classification that won the ILSVRC-2012 competition and achieved a winning top-5 test error rate of 15.3%, compared to 26.2% achieved by the second-best entry. We see the architecture and compare it with GoogleNet and ResNet

machine learning

Floating point operations per second (FLOPS) of Machine Learning models

In this article, we take a look at the FLOPs values of various machine learning models like VGG19, VGG16, GoogleNet, ResNet18, ResNet34, ResNet50, ResNet152 and others. The FLOPS range from 19.6 billion to 0.72 billion.

machine learning

Types of Activation Functions used in Machine Learning

We explored the various types of activation functions that are used in Machine Learning including Identity function, Binary Step, Sigmoid, Tanh, ReLU, Leaky ReLU and SoftMax function. Activation function help the network use the useful information and suppress the irrelevant data points

machine learning

Evolution of CNN Architectures: LeNet, AlexNet, ZFNet, GoogleNet, VGG and ResNet

It all started with LeNet in 1998 and eventually, after nearly 15 years, lead to ground breaking models winning the ImageNet Large Scale Visual Recognition Challenge which includes AlexNet in 2012, ZFNet in 2013, GoogleNet in 2014, VGG in 2014, ResNet in 2015 to ensemble of previous models in 2016.

machine learning

You only look once (YOLO) algorithm

You only look once (YOLO) is a state-of-the-art, real-time object detection system. It is a fully convolutional network. On a Pascal Titan X, it processes images at 30 FPS and has a mAP of 57.9% on COCO. It has 75 convolutional layers with skip connections and upsampling layers and no pooling.