ARTIFICIAL INTELLIGENCE, MACHINE LEARNING, DEEP LEARNING EXPLAINED AND UPCOMING APPLICATIONS IN 2019

by Nauman Jaffar, MarkiTech

In 1997, Garry Kasparov was defeated by Deep Blue, an IBM super computer. This was historic because for the first time ever a computer program had defeated a world champion in a match, which was played in accordance with the tournament regulations. This event caught the imagination of data scientists and software engineers who started exploring the yet undiscovered world of Deep Learning.

In 2015, we created MarkiTech with a focus on solving business problems using data from IoT sensors and making sense of them using Big Data, AI and Cloud computing technologies. While working for major telco, banks and management consulting companies - we also started on our journey to solve major problems using these emerging technologies and hence LocateMotion and YourDoctors.Online were born with other founding members. Our primary focus at MarkiTech was to understand the field of Data Engineering and how it will solve various business problems. This is the best picture I was able to find that helped me clarify the difference between a business, engineer and scientist in this field.

Before going deep into the AI world and taking a few courses in AI/ Big Data and with my background in IoT, Digital Transformation and app development – let me try and summarize the big picture as there is so much being written and talked with respect to the emerging technologies. This is like summarizing the meaning of life (as complicated) but this picture below shows how overlaps and various technologies come together in the financial world only.

Now let us get back to the topic in discussion. Even though Deep learning is a buzzword today, the application of this technology will not go mainstream before a few years. It has nevertheless started having an impact on our lives and will surely grow in 2019. Take one example of the Financial Industry (FI) and you will see below what I mean by that it will take some time to go mainstream.

To understand the significance of Deep Learning and its applications, we have to first learn about some fundamental concepts. The figure below shows the classification of Artificial Intelligence. As the figure depicts, Deep learning is a sub-set of Machine Learning. And Machine learning is a subset of Artificial Intelligence.

Deep learning is a subset of machine learning which in fact is part of the world of artificial intelligence

ARTIFICIAL INTELLIGENCE is what makes machines to become steadily or progressively intelligent. It is when a machine completes a set of tasks based on an algorithm or a set of given rules that solve problems.

MACHINE LEARNING means empowering computer systems with the ability to “learn”. ML is to enable machines to make accurate predictions and to learn by themselves using the data which is provided to them.

Machine learning is a subset of artificial intelligence; it is a technique for achieving Artificial intelligence. It is a method where algorithms are trained in a way so that they can learn how to make decisions.  Training, in Machine learning means giving a lot of data to the algorithm so that it can learn more about the information.

For example, here is a table that identifies the type of fruits based on its characteristics:

As shown in the table, the fruits are differentiated based on their weight and texture. However, the last row gives only the weight and texture, without the type of fruit.

Now, a Machine learning algorithm can be developed to identify the fruit. After the algorithm is provided with the training data, it will learn the differentiating characteristics of the two fruits. Therefore, if provided with data of weight and texture, it can predict accurately the type of fruit with those characteristics.

It is important to note that these Machine learning models rely on humans to provide the input features. For a model to be effective the input features need to be useful. They rely on the intuition and domain knowledge of the modeler. The modeler will have to feed the Machine learning model with the correct features. Here is another example in healthcare of machine learning where human interaction (doctor) is required to provide input to the data.

DEEP LEARNING is a subset of Machine learning; rather, it is a technique for achieving machine learning. DL algorithms are inspired by the information processing patterns found in the human brain. Just like we use our brains to identify patterns and classify various types of information, Deep learning algorithms can be trained to achieve the same tasks for machines.

The brain usually tries to process the information it receives. It does this through labelling and assigning the items into different categories. Whenever we receive a new information, the brain tries to compare it to a known item before classifying it, which is the same concept used by deep learning algorithms.

These artificial neural networks (ANNs) are a type of algorithms that try to copy the way our brains make decisions. A neural net consists of a lot of simple processing interconnected nodes. A deep neural network has three types of layers: An input layer, Hidden layers and an output Layer. Neural networks work on a simple to complex pattern recognition. They learn simple features in the first layers of the net. Some nodes are activated on defined threshold. These activated nodes provide input to subsequent layers and the process continues until it computes the final output in the output layer.

Comparing deep learning vs. machine learning can assist you to understand their subtle differences.

For example, while deep learning can automatically discover the features to be used for classification, Machine learning requires these features to be provided manually. Furthermore, in contrast to Machine Learning, Deep Learning needs high-end machines and large amounts of training data to deliver accurate results.

The progress in deep learning is attributed to three key components. Big Data, Cloud Computing, and an exponential growth of data created due to the rise of the Internet, social media and smartphone revolution.

Deep learning frameworks can carry out tasks that humans excel at. Like image recognition, speech translation and recognition. Recognizing patterns in images and identify objects. Processing languages, understanding and classifying them into intents ad entities. Deep learning networks are being used extensively in tasks such as:

Computer Vision: Application of this include identifying objects and faces in images or video streams

Natural Language Processing: Deep learning networks like Microsoft’s and Ali Baba’s speech learning models now excel in the analysis and synthesis of natural language and speech better than humans. (as demonstrated at the SQuAD challenge, Stanford Question Answering Dataset)

Applications of speech recognition include recognizing human speech, recognition of intent and entities in a conversation or text. Converting a speech to text and vice versa.

Deep learning framework is central to the rise of Artificial Intelligence. It is a developing field and its application will increase in the coming years and will continues to transform our lives and the world we live in.

Finally I would to close out this well researched write up with a mention of MarkiTech and what we provide from our brand new 1 University Av, Toronto, AI Lab office where we also work on YourDoctors.Online and LocateMotion which are Big Data / AI / IoT focused initiatives.

Come visit us or send a note to me at [email protected];

References:

Deep Learning, Goodfellow, Bengio and Courville, 2016

MGI-Artificial Intelligence: The Next Digital Frontier

Wired: AI BEAT HUMANS AT READING! MAYBE NOT

Clearing the Confusion: AI vs Machine Learning vs Deep Learning Differences, Dr. Michael J. Garbade

#AI #BigData #IoT #DigitalTransformation in downtown Toronto