Artificial Intelligence ApplicationsEstimated Reading Time: just 5 min

The definition of Artificial Intelligence (AI) is the intelligence demonstrated by machines, in contrast to the natural intelligence displayed by humans and other animals.

Acting like a human is defined as being able to perform 5 macro-functions:

  • Perception (recognizing and interpreting reality)
  • Reasoning (drawing new conclusions)
  • Planning (detecting patterns and adapting to new environments)
  • Knowledge Representation (extract and store knowledge)
  • Natural Language Processing (communicate successfully in a human language)

AI research started in 1956 after a workshop at Dartmouth College and focused on solving only one of these 5 traits at a time. This is called Restricted AI.

Through the years AI research was able to integrate more macro-functions starting the General AI era.

In detail, two things changed everything in AI:

  1. Machine Learning that uses statistical techniques to give computer systems the ability to learn from data, without being explicitly programmed
  2. Deep Learning that is a subset of Machine Learning which allows machines to learn a large variety of activities as humans do

Nowadays, there are 3 main problems that can be solved with AI

  1. Regression Problems: this could be applied whenever the output of a subset of inputs is a continuous function. The way for determining such a function is using Supervised/Semi-Supervised/Unsupervised Learning so that the machine can forecast the outcome.
  2. Classification Problems: this could be applied whenever the output of a subset of inputs is a discrete function. The way for determining such a function is using Supervised/Semi-Supervised/Unsupervised Learning so that the machine can classify the inputs (i.e. 1/0, car/human/bicycle).
  3. Clustering Problems: this could be applied whenever the output of a subset of inputs is not predictable/known. The way for determining such a function is using Supervised/Semi-Supervised/Unsupervised Learning so that the machine can cluster the outcome.

These problems are the basis for machines to take decisions.

The 3 problems are solved by using sequences of algorithms and every algorithm is called layer.

The subset of layers needed for performing a task is called Neural Network.

Usually, every Neural Network performs 4 activities:

  1. Data Collection
  2. Data Cleaning
  3. Patterns Identification
  4. Prediction Making

To understand how this works, let’s think about the job of a machine solving a Classification Problem such as recognizing the picture of a dog.

First, the machine must learn what a dog looks like.

This may involve the following steps repeated several times:

  1. Seeing a picture (Data Collection): this means translating the picture into a binary code
  2. Rotating the picture (Data Cleaning): this could mean recognizing the eyes and aligning them
  3. Screening the rotated picture (Patterns Identification): this means translating the rotated picture into a binary code (this is different from the one in step A since the picture has been rotated)
  4. Learning if it is a picture of a dog (Prediction Making): this may happen by
    • Getting a confirmation by a human (Supervised Learning)
    • Comparing the output of step C with an archetype of a dog that the machine built in binary code (Unsupervised Learning)
    • A mix of the previous cases where the machine expresses a prediction together with the related degree of accuracy of it and a human gives/confirms the right answer (Semi-supervised Learning)

A machine will be able to know if a picture it is looking is a picture of a dog when it will have analysed a number of images that makes a relevant statistical sample for it to make a prediction with a fixed degree of accuracy.

When a machine knows what a dog is, it will be able to discern if it is seeing a picture of a dog or not.

This may involve the following steps:

  1. Seeing the picture (Data Collection): this means translating the picture into binary code
  2. Rotating the picture (Data Cleaning): this means recognizing eyes and aligning them
  3. Screening the rotated picture (Patterns Identification): this means translating it into binary code
  4. Understanding if it is a picture of a dog (Prediction Making): comparing the output of step C with an archetype of a dog that the machine learned and stored in binary code

In the example:

  • Machine Learning solves a Classification Problem to create the archetype of a dog
  • Neural Network is the application of sequential algorithms performing the 4 steps

Nowadays, the main applications for AI are

1. Regression Problems

  • Financial Forecasts such as Stocks Market Trends, Investments and Fraud Detection
  • Demand Forecast (i.e. Heineken)

2. Classification Problems

  • Scene Understanding (i.e Self-Driving Cars)
  • Recommendation Systems (i.e. Thetake.ai, Facebook Pixel, Pinterest and Netflix)
  • Disease Detection
  • Defect Detection
  • Predictive Maintenance
  • Language/Speech/Text Processing (i.e. x.ai)

3. Clustering Problems

  • Market Segmentation (i.e. Energy Consumption by OPower)
  • Root Causes of Problems Determination

BIG DATA ANALYTICS

Big Data Analytics, commonly called Big Data, is the inductive process of examining data sets to uncover information such as

  • patterns
  • correlations
  • trends
  • preferences

to make informed business decisions and forecasts.

The data sets used for Big Data are composed by both:

  • quantitative data
  • qualitative data

The most important thing is that Big Data works with both

  • structured data
  • non-structured data

This means that, differently from the classic Business Intelligence, Big Data can upload and examine databases made by both

  • numbers
  • contents

as well as pulling data from

  • corporate datasets
  • external datasets

This means that Big Data allows, for instance, making predictions about sales using both revenues data and users’ activity on social networks.

What makes this possible is the use of Artificial Intelligence and, in particular, of Deep Learning.

Nowadays, there are 3 main problems that can be solved with AI

  1. Regression Problems: this could be applied whenever the output of a subset of inputs is a continuous function. The way for determining such a function is using Supervised/Semi-Supervised/Unsupervised Learning so that the machine can forecast the outcome.
  2. Classification Problems: this could be applied whenever the output of a subset of inputs is a discrete function. The way for determining such a function is using Supervised/Semi-Supervised/Unsupervised Learning so that the machine can classify the inputs (i.e. 1/0, car/human/bicycle).
  3. Clustering Problems: this could be applied whenever the output of a subset of inputs is not predictable/known. The way for determining such a function is using Supervised/Semi-Supervised/Unsupervised Learning so that the machine can cluster the outcome.

The 3 problems are solved by using sequences of algorithms and every algorithm is called layer.

The subset of layers needed for performing a task is called Neural Network.

Usually, every Neural Network performs 4 activities:

  1. Data Collection
  2. Data Cleaning
  3. Patterns Identification
  4. Prediction Making

To better understand how Big Data works let’s walk through a process step by step.

STEP 1. DATA COLLECTION

The sources for Big Data Analytics can be several but the most important are:

  1. Corporate Datasets
    • ERP
    • CRM
    • Databases (i.e. Sales, Marketing, Finance, Planning Databases)
  2. External Datasets
    • Sensors (i.e. the ones present/installed on a machine)
    • Social Networks
    • Videos/Photos
    • Documents (i.e. a pdf or an Office file)
    • Web Data

The acquisition is made through the part on Neural Networks that perform Data Collection and Data Cleaning. When qualitative/non-structured data are collected, Vision Systems are often involved.

The output of the Data Collection Step is called Raw Data.

STEP 2. DATA ANALYSIS

The input for this step is Raw Data collected in the previous stage.

Raw Data is then processed by Neural Networks that perform Patterns Identification and Prediction Making.

The analysis that can be performed are:

  • Predictive Analytics
  • Clustering Analysis

Predictive Analytics elaborates Raw Data to make predictions about the future through Regression or Classification algorithms.

Clustering Analysis elaborates Raw Data to group a set of objects in such a way that the ones in the same group are more similar.

The output of Big Data is a Front-end that often performs real-time reporting through

  • Visualization Tools
  • Online Business Apps

BUSINESS CASES

Now that you learned everything about Artificial Intelligence, would you like to go through some real cases?

If the answer is yes, follow this link!

Nicola Zaffonato Administrator

Business Strategy | Product Marketing | Executive Master eCommerce Management | Business Innovation Master | MSc

I am driven by my personal growth and of people/contexts that surround me.

I followed a professional path in Valentino Fashion Group and Luxottica during which, thanks to the ability to understand different businesses and interests, I was able to succeed in Operations, Merchandising and Retail.

These organizations have exploited my ability to mediate and translate needs/constraints into practice, assigning me to Project Management roles.

Luxottica relied on my ability to analyze, to anticipate things and to imagine/implement solutions by appointing me in Supply Chain Management department and assigning me to the Product Management of IoT solutions for Anti-counterfeiting and Retail digitalization.

During this professional path, I also developed my leadership by managing teams to build Processes, Organizations, Systems and Governance Tools.

follow me
Sharing is Caring!
en_USEnglish
search previous next tag category expand menu location phone mail time cart zoom edit close