Featured

Automation, Intelligent Applications, Things and the future of business.

This is the post excerpt.

Machine Learning is changing the world.

Not long ago, when an application(enterprise or otherwise) was deployed, it was considered a black box. No one knew what was happening, and log files were cryptic entities that were looked into only when users complained about issues.

Fast forward to the current day, and we have intelligent algorithms telling us what the users did(and did not do), recommending offers that are suitable for individual users. We also have some adventurous companies “instrumenting” the real world, using sensors, and segmenting the customers and employees with data gathered from the environment.

Gone are the days when companies were left making best guesses about the “success” or “failure” of a campaign. In these days of A/B testing, users do not need to rely on opinions, when the decision can be based on cold, hard data.

Also gone are the days when decisions were taken by HiPPOs (“highest paid person’s opinion”). These days even the junior most employee can put his ideas to test and expect a proper data-based answer in a matter of hours or days.

What kind of applications should we build in this brave new world?

Some examples are:

  • Applications that can adapt based on user’s action or inaction.
  • Applications that can respond to stimuli in the environment(both virtual and physical)
  • Applications that recommend actions based on the goals specified.
  • Applications that can continuously learn and improve/fix itself via. multiple ways like anomaly detection, user actions, signals from the physical world etc.

Deep Learning (or what is so deep about it?)

If you are a techie, or follow technology blogs or news, the odds are that you will have heard about Google’s AlphaGo AI beating the world champion.

That was in early 2016. Since then Deep Learning(the technology behind AlphaGo) has made significant advances.

Deep Learning has transformed businesses and also enabled brand new products and services.

Some industries/areas that Deep Learning has impacted:

Healthcare : DL based systems are now better at reading X-rays than human doctors

Education: Personalized education is now possible because of DL, and this revolutionizing the educational field.

Agriculture : Precision Agriculture is another area where DL has had an impact.

Self driving cars: is currently the favorite example given by DL practitioners (when they are asked about exciting areas where DL is being used.)

Some of the other areas are detecting spam in emails, recognizing images in a picture, detecting fraud (and thus minimizing risk) etc.

Almost any business problem you can think of, is being solved by innovative companies across the world using this cutting edge technology.

For example, have you wondered about any of the following business questions:

  • Understanding your customer’s intent.
  • Desire to provide a great customer experience that positively impacts your core brand value.
  • Want to implement dynamic pricing, depending on multiple real-time factors(like product demand, brand awareness, environmental and other factors)?
  • Desire a ‘no unplanned downtime’ standard for your equipment/machines?
  • Want to find out the risk portfolio that is optimal for you?

So what is Deep Learning and how does it work?

Deep Learning is loosely based on the structure of the human brain, and Neural Networks are at the heart of this technology.

DL works by utilizing the massive data that companies already possess (think customer data, call logs, emails, support tickets, security videos, transaction information, application logs, supplier data, ‘iot’ data from your machines and sensors) and turning them into actionable insights…Companies use DL to predict when machines/equipment will fail (predictive maintenance) and others use DL to detect human emotions on customer’s faces when using their products or services(customer satisfaction).

In future posts we will look more closely at :

  • Neural Networks and how to build and train a deep NN.
  • How to optimize your NN so that it performs well…In particular we will look at:
    • Hyper-parameter tuning,
    • Regularization and
    • Advanced Optimization Algorithms like Adam Optimization Algorithm
  • Look at CNNs(convolutional Neural Networks) and use them to build models that can be used to classify images (and also videos).
  • Look at Sequence Models like RNNs and LSTM models