Data science is an inventive interdisciplinary field that involves mathematics, computer science, modeling, statistics and analytics. Innotomy can assist in

* Machine Learning

* Predictive Analytics

* Deep Learning

* Computer Vision, Graphics and Image Processing

* Medical Imaging

* Biochemistry, Molecular Biology and Genetics

* Bioinformatics and Genomics

* Energy, Water and Air

* Communications and Signal Processing

* Quantum Computing and Quantum Cryptography

* Applications of Blockchain

* Astronomy

Machine learning is a branch of AI that explores way to get computers to improve their performance based on experience. The popular techniques used are:

* Supervised Learning - used to find a mapping function for various inputs/outputs.

* Unsupervised Learning - Used for clustering

* Reinforced Learning - Uses best past experience

* Transfer Learning - transfer learning from one deep network to another

Predictive Analytics is used to make predictions about unknown future events by modeling the known past events. It draws many techniques from statistics, modeling, machine learning and artificial intelligence. It uses among others

* Linear Regression and correlation

* Logistic Regression

* Decision Tree Learning

* Random Forests, MARS, Clustering

* Naive Bayes Classifier

* KNN, SVM and Gradient Boosting

Unlike shallow learning algorithms used in predictive analytics, deep learning involves transforming the inputs through a series of many (hidden) layers, at each layer, the parameters of its processing unit are learned through training. The well known constructs in deep learning that use backpropagation algorithm are:

* Artificial Neural Networks - SP, MLP

* Deep Neural Networks - LSTM, RNN and CNN

* Theory of Information Bottleneck

Deep Neural Network architectures are used for solving specific problems:

* Restricted Boltzmann Machines

* Convnet - Convolutional Neural Networks

* LeNet - Python library for ConvNet (GoogLeNet)

* AlexNet - ImageNet classification with deep networks

* ResNet - Deep Residual networks

* R-CNN - Region based CNN in deep networks

* RPN - Region Proposal Networks

* FPN - Feature Pyramid Networks

* Kaggle - A platform for competing

Cloud Computing has transformed the data science industry by making high end computations accessible to everyone at very low cost. Amazon Web Services and Google clouds are two examples.

* A single NVIDIA GPU makes a DNN run 30x faster

* Pay per use business model

* AWS

* Google Cloud

* FloydHub

* Paperspace

The languages that make data science happen!

* Python - By far the most popular. Numpy, Scipy and Pandas are enhancers

* R - Preferred language generally in Genomics. Powerfully backed by Bioconductor

* Julia - A language for high performance numerical analysis and computational science, also used as spec language in quantum computing

Deep Learning libraries for Python have made it possible to implement and solve many learning tasks with ease. The outstanding ones are:

* Keras - Deep learning library for Theano / TensorFlow

* TensorFlow - Google backed mathematical backend with a simplified high level abstraction

* Theano - Mathematical framework that combines power of CAS with optimizing compiler

* CNTK - Microsoft's Cognitive Toolkit

* Torch - Open source machine learning library