Hello friends, welcome back. In the previous session, we discussed the fundamentals of cognitive computing. We understood what can make a system cognitive, what are all the uses of cognitive computing, application sector and more things we have discussed and I hope it would have been interesting. Now in this session, we are going to go into the next level where we will understand applications in depth, what makes a system real cognitive stuff and what are all the attributes for the cognitive system. will all be discussed very clearly in this session.
We have been hearing the terms machine learning, speech recognition, computer vision, natural language processing, all these for quite a long time. Are these cognitive? Answer is a big yes. Many examples are there which we are already using or are already falling into the cognitive systems. They are categorized as cognitive systems without me understanding that we are using them as cognitive systems.
So let's understand some of the examples. of cognitive systems which are already being handled by us and which we are very familiar with. The first one is natural language processing.
We know what it is. It enables your computers to understand and interpret human language as they speak and it is making it possible for the machines to interact in a very natural way through return or spoken language. The slang that you use, the way you speak, the way you present data, the way you type, it doesn't matter.
It will give you a reply. and that's natural language processing for you. Machine learning, we know what it is. Machine learning algorithms allow systems to learn from the data.
The data is provided by you and it will improve their performance over time without being explicitly programmed. Human intervention will be reduced and it's going to learn from the data. It is going to be completely nurtured by the data and it will also improve the performance over time. I have also defined earlier in my machine learning playlist, clearly the machine learning definition, deep learning definition, difference between machine learning and deep learning. All those have been clearly explained to you and I request you to go through it if you are not familiar with it.
The next one that we have been already using for ages now is speech recognition. Cognitive computing systems can easily recognize and interpret human speech enabling voice activated applications and virtual assistants. Your google assistant is the best example.
We have been using it day in and day out. and that's the best example for speech recognition. They come and fall under the cognitive computing as well.
The next one is computer vision. The computer vision technologies actually helps computers to interpret and understand visual information from images or videos. One simple example would be you are going into a supermarket.
You want to know what are all the products in the rack. You have an app. The app is launched. The app understands what are all the products available in the rack or the shelf through video. and it tells you what is the product, what is the cost.
You are waiting in the road for the signal, your car automatically recognizes the signal and it gives you update about you can move or not and this is happening through the video recognition and that comes under computer vision. We have been using computer vision in many ways and that already has come under the cognitive systems. Recommendation systems. The moment you open any app like Amazon or Flipkart or anything, you are getting product recommendations.
based on your historical purchases, based on your search interest, all those things, they also fall under cognitive systems. You are basically provided with an input where the input is going to be created or input is going to be curated through your previous interests. That is called recommendation. For example, I am reading news consistently about cricket and if the cricket related news is popping up, that's through the recommendation system and it understands clearly that I am a cricket fan. Problem solving and decision making is another area which is 100% cognitive.
Cognitive systems can really analyze the complex data and it can consider the different scenarios probabilities and it can get into a very good informed decision. You will get decisions created, you will get decisions curated through proper analysis. So NLP, machine learning, speech recognition, computer vision, recommendation system, problem solving which includes decision making.
are all cognitive in nature. We are already used to this and we know this. Just a thing, we are recollecting it here.
Please remember this point. I have been talking about cognitive computing definition for quite a long time now. We need to recollect. The cognitive computing is not here to replace the human intelligence. It cannot and no technology can.
But it is going to augment it. It is going to support it. It's going to provide valuable insights, support and automate tasks that are repetitive in nature and which can reduce the human workload and pressure on us. So, The cognitive computing or any other computing for that matter is not going to replace human. It's going to support.
It's going to augment. The field of AI is continuously emerging. It's going to be emerging further as well. Cognitive computing is expected to play a huge role, vital role in shaping up the future of technology as a whole and also the impact is going to be really very good on the society.
So cognitive computing cannot replace us. It can augment us. Remember this point. This is a very strong point that I would like to stress upon. Any technology can support us and we cannot be replaced.
Our brain is the most powerful. Application perspective is very important. We need to understand where all cognitive systems can play a major role. Let's take one by one and these are all already proven and some of them are evolving as well.
You can identify your own scenario. and you can build applications for a particular domain as well which is most welcome and that's what is needed. As I have already conveyed you, cognitive computing is one area which is evolving. It has a lot more in the store for all of us to utilize and it's time that we can wake up and we can build more things around cognitive computing and it can be really really helpful for us. Let's start with healthcare.
Cognitive computing has played a major role in the healthcare. Many systems have been already built and it is proven vital. It can help you in diagnosis, it can help you in treatment planning, it can help you in analyzing the patient data, the medical records and research literature.
Most importantly, cognitive computing has been so much useful in the drug discovery and personalized medicine and healthcare management. People have a lot of cognitive computing impact in this medicine area and many articles have come in this area saying that this is going to be a boom and this is already supported, this is already presented us with a lot of useful applications. Customer service is the next area I'm going to touch. As we all know, gaining customers is not tough, but retaining them is very difficult and serving customers should be proper and it should be on time.
Cognitive computing is used to build extremely powerful chatbots and virtual assistants which can handle the customers. The customers'queries, the customers'expectation in terms of product recommendations and customers'expectations in terms of immediate reply. The customers are nobody is expecting us to give reply immediately. So all those can be met through the cognitive computing based solutions.
Next comes finance. Yes, in the financial industry, cognitive computing is playing a vital role. It can help us in detecting the fraud, the risk assessment. I am going to give somebody a loan.
Will he be in the ability to repay the loan? Can he repay the loan? Is this person trustworthy? All these kind of things can be analyzed. clearly using cognitive computing.
The algorithmic grading, personalized financial advice. I have this much money with me. I need to invest it in this many areas I believe but is it a right decision? I can use cognitive computing to understand what I'm thinking is right or may not work out. So these kind of personalized inputs with respect to financial advice is also made possible through cognitive computing systems.
Education. This is the most important sector. Cognitive computing can definitely improve.
your educational experiences and it can provide personalized learning path for the students or the faculty members and most importantly the pain point with respect to grading correcting papers can all be corrected can all be rectified and cognitive computing can play a major role there it can create interactive and adaptive learning materials as well teaching is difficult learning is very very difficult we have to make it interactive and that's what cognitive computing is going to provide us in near future Some applications have already come where it has made the learning and development very easy and very interactive. NLP, I highlighted this sometime back, NLP including the natural language translation, sentiment analysis, voice recognition, text summarization and more applications which enables easier interaction between humans and machines possible. All these are built through NLP and it is nothing but coming under cognitive computing. Autonomous vehicles.
Cognitive computing play a vital role in the autonomous vehicles by enabling it to take real-time decisions. Do I have to take a left or right? There is a heavy congestion, where do I move?
What can be the speed that I can go in? There can be a virtual bumper that can come in front of you if you are driving rash. It can assist you in moving from one lane to another. It can let you know if the signal has changed from green to red.
All these things can be done and it can analyze traffic patterns real-time and it can assist you. It is not only to monitor the drivers we are building systems, we also build the systems to support and aid drivers. Not only that, we can also give comfort and safety as a paramount feature. to the passengers through cognitive systems.
All these are in place and many advanced automobiles have used ADAS, Advanced Driver Assisted System and DMS, Driver Monitoring System in place to support this. So we have application perspective discussed right now and it is not going to end here. We have got lot of applications coming in picture and we are going to discuss the next set of applications here in this slide.
Manufacturing is another sector where we can really optimize the manufacturing process. and most importantly, prognosticative approach. You can go ahead and analyze, predict the failures even before the failure occurs. I am having a huge machine.
If it fails, the entire productivity will fail, entire floor shop, job shop will be shut. Before that, if I use cognitive computing, I will be able to even predict these failures and improve the entire chain. Marketing and advertising has got a huge, huge application possibilities with the cognitive computing. It can analyze the consumer behavior. this customer when buying a product how happy he is and what kind of other options he is having in his mind what is his preferences how do we analyze the market trends when do we introduce a particular product has it are we going to introduce it in the seasonal basis can i introduce this particular product during diwali during ramzan during christmas what do i do i get all those kind of insights through the marketing and advertising area which i can build which i can build solutions for this area using cognitive computing Gaming and entertainment, this is very interesting for all of us.
We can definitely create more intelligent and responsive characters as well and it can be more more interactive for the players to play. The virtual reality experiences can really be enhanced using the gaming and entertainment options that we can use with the cognitive computing to build solutions around it. Security, yes, cognitive computing plays a real vital role in the security sector as well.
It can detect. the cyber security threats it can identify the patterns and it can let you know if it is suspicious and it can protect sensitive data believe me wherever you are going you can take cognitive computing alongside you and it can build solutions for you agriculture the paramount area where definitely cognitive computing has a huge say and a play it can go ahead and help you with precision agriculture it can help you in optimizing crop yields it can monitor livestock health and most importantly it can even predict if this particular crop is going to be affected by insects or are they going to be kind of troubled through the environmental situations all around us and what kind of solution can we give for that. It can help in predicting the crop health.
It can suggest you what kind of fertilizers can you use, what kind of approaches can you use. So agriculture has also got a huge huge opportunity and we can use photographic computing there. And finally last but not the least research and data analysis.
use the cognitive computing to analyze and handle real huge voluminous data sets and discovering patterns and understanding insights from those patterns is really important and you are going to do that through cognitive computing options and astronomy or climate science has got a huge amount of data every time when you want to analyze it. It's not easy but we can make it possible through cognitive computing. So we discussed the application sectors which started from healthcare, customer service, finance, education, NLP, autonomous vehicles, manufacturing, marketing and advertising domains, gaming, security, agriculture, research and data analysis area.
These are all the sectors that we could identify and there are definitely more to it and I'm sure you can explore a little more as well. So all these are interesting areas where you must be already working and I believe that these insights will help you in getting it better. What makes a system cognitive? I have a system, what if I add can make it cognitive.
That's a very important topic for discussion and we are going to take it right now. There are three important things which I have already conveyed you very clearly. One is learning.
I need to learn consistently if I call myself cognitive and you need to learn data across the time. Second is I need to go ahead and generate insights through the model that I'm building. So model building is the second step.
Third step is hypothesis generation. I explained you these three points clearly in my first session. So learn, build model, generate hypothesis. These things will make the system cognitive. But is that the end of it?
No. I've got a better view where we are going to understand what all in addition can make it real supercognitive. The system is considered cognitive when it possesses certain key attributes that can mimic, I repeat, that can mimic human-like cognitive abilities. These attributes enable the system to understand, reason, learn and interact with users in a more natural and intelligent manner.
This is the definition I reiterated here because we need the system to understand, reason, learn and interact. When you do all this, it is cognitive. Well, let's list the key factors which are very important for a system to become cognitive.
First is NLP. We have been using this term NLP, NLP, NLP from the beginning to understand the importance of it being used for building a cognitive system. Cognitive system can process and understand natural language.
It can allow the users to interact with the system, communicate with the system in a real natural way and more human-like way. When you talk about NLP, it can be speech recognition, language understanding, language generation, can be anything and you need to understand if a system is getting cognitive, NLP is going to be a component there and this is one of the key factors. Learning and adaptation. Cognitive systems, as I already told you, must learn. It cannot be reluctant.
It cannot say that I know all these, that's all done. It cannot ever say that I am done with my learning. The learning process is going to evolve and it should learn consistently and continuously from the data and the experiences and that will enable them to improve the performances over the time without the programmer touching the code explicitly.
The learning ability is very very important for the cognitive systems to get in into the new situations and make more informed decisions. This is a very important aspect of how a system can become cognitive. Reasoning and problem solving, I highlighted this as one of the very important things some time back.
A cognitive system must be capable of analyzing the complex data and it should derive the inferences out of it and it should also go ahead and apply reasoning to it and make decisions based on the insights derived from it, right? Multiple factors are to be considered. It is not easy.
and a cognitive system will be capable of doing this. And context awareness is very, very important, right? You have to interpret the information based on the circumstances and the situations. The context is very important when you are talking about cognitive system. On what context is this user talking?
It can be sarcasm, it can be happiness. Wow, what a product. This can be real happiness or this could be sarcasm.
So the context needs to be understood very clearly when you are building cognitive system. The cognitive system should also be very interactive and engaging in nature. If they are not interactive, they cannot be cognitive.
If they are not easy to use, engaging to use, they cannot be cognitive. It should be more like human-like conversational experience that we should provide rather than thinking about a machine-like experience and perception and sensing. Cognitive systems must be capable of interpreting the sensory data, interpreting and understanding the data from different sources.
The sources can be images, it can be videos, it can be sensor inputs. Irrespective of what it is, they should be capable of getting it done. And knowledge representation is the next thing that we need to understand. Cognitive systems can store and organize knowledge in a structured format. Understand, it can store and organize.
The part organize is very very important and that enables them to access and to use information whenever needed. I am going to use the data day after tomorrow. or next month or next year but when I go and use the data it is in a messy way I cannot use it. Cognitive system helps you in getting it arranged in a real nice pattern structured way and whenever you need that can be retroactive. Emotional intelligence is also expected as one of the key factors in some cases because you are going to talk to you are going to interact to humans.
When you are interacting with humans emotional intelligence is going to be always helpful and when you recognize and understand the emotions of the human It's going to be really helpful to give a reply in an empathetic manner. Empathy is going to be added to all these kind of products that come with cognitive intelligence. Cognitive systems are to be empathetic. When you are interacting with humans, you are to be empathetic and that's what is stressed here. What are the common attributes for cognitive system?
Learn. You must learn from the experience, from the previous data, from the evidence and improve your knowledge. If you are calling yourself cognitive, if your system has to be cognitive, It has to learn from the experience and the data that you are providing it as a feed and it should evolve.
Next is generate. It should be able to generate or evaluate very clearly hypothesis based on current state of its knowledge. We need to generate the hypothesis.
That's very very important. Report on findings is the next step that we need to do. You have learned, you've generated the hypothesis, go ahead and report the findings in such a way that it can justify the conclusions clearly based on the confidence the evidences that you have obtained then Discover.
Discover the patterns in the data. You can take guidance from the user or you need not but the patterns will be really helpful and this is the next step. What did you do in the beginning?
You learned. Then what did you do? You generated hypothesis.
Then you have reported the findings and then you are discovering the patterns. Now you are going to emulate. You are going to emulate the process or the structures found in the natural learning systems and this is very very important step. the memory management, knowledge organization, modeling the brain structures, processes, all these things can be emulated.
You can emulate all this stuff and that's going to be really helpful. And you can use NLP to extract the meaning, the insights from the textual data, and you can use deep learning to extract features from images, video, voice, or sensory field. And you can deploy variety of predictive analytics algorithms and statistical techniques clearly.
So, these are all the attributes, the common attributes that the cognitive systems must have. Learn, generate, report, discover, emulate, use, deploy. I hope it was clear and understandable.
In case you have any questions, you can always ping me in the chat. I'll be very happy to clarify. Thank you very much for following my channel content.
If you have any questions, please feel free to ask me. Thank you.