Innovating Industry 4.0 with machine learning

About Event

Industry 4.0 is a paradigm shift in the way the manufacturing industry works by inculcating technologies like industrial internet of things (IIoT), machine learning, and artificial intelligence. The increasing demand for customized products at reasonable rates is the principal driving force behind the need to use various aspects of AI and machine learning in the manufacturing process.

Machine learning addresses key challenges like unexpected equipment downtime, wasted labor, and production losses. This event is aimed to detail the role of machine learning in the growth of Industry 4.0. Our subject matter experts will brief you on machine learning and explain the basics of how it works. They will also discuss how this technology impacts various areas of operations of different industries.

Agenda

• Introduction of Industry 4.0
• Importance of Machine learning
• Real world implementation of ML
• Supervised and unsupervised ML implementations
• ML opportunities in various industry
• Industry specific ML use cases
• Q & A

Questions & Answers

The following are the answers to the questions that were asked during the event.

Question 1: What tools and technology you use for ML and DL?

Answer 1: We are mainly using following technologies:

• Tensor flow + Keras with Nvidia GPU for on-premise solutions and Azure VM for cloud Deep learning
• Scikit-learn and PySpark cluster for machine learning for on-premise and cloud

Question 2: Can you give an example of situation awareness?

Answer 2: In one of our major projects, we had implemented smart mirror solution where advertisements were shown based on human face / emotions. It was mainly implemented for our client to run smart campaigns on latest products and learn how users / shoppers behave through their emotions for specific products.

Question 3: How does machine data translate into data science models?

Answer 3: Translation means checking, cleaning, and preparing the data for the required model. It also includes finding the features from the data and providing them to data science model for further evaluation. There are several data management tools like D3.js, SAS, Dataflow, Informatica, etc., to provide a managed and scalable solution for putting models into production.

Question 4: How much data you need from Machine in terms of time to make an accurate model, how much accuracy does the model have

Answer 4: It entirely depends on how the data is; more data would provide better accuracy to the model. Sometimes thousands of images will be required for a model to be trained and at times historic data of 1-2 years will be sufficient to train the model. With these datasets we can get 80-90% accuracy.