AI Journey

Artificial Intelligence: Any technology which enables computers to mimic human behavior in 1950’s to 1980’s.

Machine learning : AI techniques that give computers that ability to learn without being explicitly programmed to do so in 1980’s to 2010.

Deep learning : A subset of ML which make the computation of multi-layer neural networks feasible.

                           Why Deep Learning?

When the amount of data is increased ,machine learning techniques are insufficient terms of performance and deep learning gives better performance like accuracy.

                                                      Deep Learning vs Machine Learning

Deep neural networks are biological inspired learning networks which are good at extracting information from unstructured  data without well designed features.

Multilayer Perceptron: It is a supervised learning model trains weights across its hidden layers by reducing the prediction error through back-propagation.

Convolutional Neural Networks(CNN):

1. It is used in image classification and when there is a spatial pattern to the input data.

2. A CNN model scans through the images to learn presence of particular features such as curves or edges.

Convolutional Layer:

1. A CNN is a neural network with some convolutions layers and some other layers.

2. A convolutional layer has a number of filters that does convolutional operation.

Sub sampling pixels will not change the object. We subsample the pixels to make image smaller.

                            Deep Learning Application

Deep neural networks for improved traditional algorithms

1. Enhanced Fraud detection through identification of more complex problems.

2. Enhanced identification of defects based on deeper anomaly detection.

3. In store activity analysis of video

4. Satellite images: labelling terrain, classifying objects.

5. Automotive: Autonomous vehicles navigation by recognition of roadways and obstacles.

6. Healthcare: Diagnostic opportunities from CT, MRI .

7. Insurance: Estimating claim severity based on photographs.

8. Customer satisfaction: Transcription of voice data to text for NLP analysis.

9. Social Media: Real-time translation of social and product forum posts

10. Photo captioning: search archives of images for new insights.

11. Finance: predicting behavior based via time series analysis.