“No invention happens without intelligence. “
“Virtually personified human mind is what artificial intelligence is”
Back again with a new age learning.
Have u guys watched a web series ‘100’. Do watch and do read following article.
When I was young my Mother used to quote the following words,” the capacity to learn is a gift and willingness to learn is a choice. “These words were imbibed in me and the affect is that I have never stopped doing so.
Each new day, moment, brings with it a new lesson. In this enthralling world today, if you want to stay ahead, learning and updating yourself with new technology has become a necessity. New age empowerment, your cell phones, laptops has become your strength and the most reliable sources. Even in the time of Covid-19, one thing of survival was the networking systems, the sensor-operated systems.
As we have come up with the word ‘sensor-operated ‘, I would highlight the three magical systems which have become the need of today: ‘artificial intelligence, machine learning and deep learning. ‘
What are these and it should be left just for geeks?
AI and Machine learning (ML) are the two most talked about buzzwords today. They open a vivid world of opportunities.AI is beyond and ahead in every niche, it has stepped in. we are in a age, where the youth is more into self-auto-powered things like Siri, Alexa, google, etc. They are least keen about doing things on their own. So, the reason behind getting acquainted with these technologies is,
- The skill of the century
- It is everywhere
- It is versatile
Bang on: Introduction
“AI brings with it a promise of genuine human-to-machine interaction. Computer science defines AI research as the study of ‘intelligent agent’, a device that perceives its environment and takes actions that maximize its chance of successfully achieving its goals. “
We always think we can solve problems on our own, but in recent studies, we have come across that computer has become more versatile and humane in terms of solving problems, both emotionally and manually with an advance in AI.
AI was first used in 1956, at a computer science conference in Dartmouth. AI described an attempt to model how the human brain works and, based on this knowledge create more advanced computers. Since then, more than 60 years, the scientist and researchers have taken many brainstorming sessions. They collected the brightest minds of the centuries and fed the computers with that. Though it turned out to be complicated, but they came out with the understanding and developed a protocol of learning, natural language processing and creativity through AI, and thus liberated humanity from many of its troubles. Though AI is ubiquitous computer is still far from perfection. But, ‘never give up’ attitude of the programmers is leading us to achieve and activate human brain into the computer memory.
Machine learning as the Wikipedia says” is a subset of AI that provides systems the ability to automatically learn and improve from experience without being explicitly programmed. It uses different algorithms that help to solve problems. “
HAs we all know, various machines nowadays work on verbal commands, scan pictures, drive automated cars and play games. It won’t be far when these will start walking among us. To make it easier to understand, I would like to place an example for you from daily life-choosing a song to play(imusic)it will ask you your liking of genre, find various options will filter and sort and finally play your choice. All this happens with the help of machine learning, i. e. computer learns on its own and provide you with a solution.
The more advanced version of AI is deep learning. We carry it in our hands most of the time: the cell 📱. “deep learning is a class of algorithms inspired by the structure of a human brain .it uses complex multi-layered neural network just like the structure of human brain, Where the level of abstraction increases gradually by nonlinear transformation of input data.
Types of AI:
- Weak or narrow AI
- Strong AI
- Super intelligent
Weak or narrow AI:
You must have heard of Deep Blue, the first computer to defeat a human in chess, and that too, Gary Kasporav in 1996.how did this happen? Deep Blue was capable of generating and evaluating about 200 million chess positions per second and is an early example of weak AI.it is widely used nowadays in science, business and healthcare. Another example is AlphaGo.
Who doesn’t like to be independent? Strong AI is the same. It is the point in the future where machines become human like. They decide and learn without any human inputs. They are competent, solve logical tasks and have emotions too.
Now the question arises: how to build a living machine? You must have watched many robotic movies, in which the machines show emotions but at the end, they have their limitations and ultimately fail the human emotions. The advancement of technology has given us chatbots and virtual assistants that are good at maintaining a conversation. Experiments are still in action to make them humane but reproducing emotional reactions doesn’t make them emotional, will they?
It is the near future of machines that are par excellence? A big question
Machines can be smart, wise, creative with social skills, and the goal may be to better the lives of human but creating autonomous emotional machines like the bicentennial Man, can be a dream come true for any of the scientists.
Right now, to make super intelligent machines they are focusing more on:
- Machine reasoning: information on database or library, includes SEO planning.
- Robotics: building, development and controlling robots.
- Machine learning: study of algorithms.
Components of machine learning:
- Datasets: the special collection of samples are called datasets. The samples can be numbers, images, texts or any other kind of data.
- Features: it is the key to the solution of the task. They demonstrate to the machine what to pay attention to.
- Algorithm: it is an ensemble learning, where we use different algorithms to achieve better performance. Depending on the algorithm, the accuracy or speed of getting the results can be different.
Any software using machine learning is independent than manually encoded instructions for performing specific tasks. If the quality of the dataset was high, and the features were chosen right, a machine learning powered system can become better at a given task than humans.
Four groups of machine learning algorithm:
- Supervised learning:
As the term reflects, in supervised learning, the trainer helps the program throughout the training process, with a labelled data training set.
Just as you teach a child in the nursery, by showing what is what, similarly, the computer runs a program on a validation set and checks whether the learned function is correct. The program makes assertion and is corrected by the programmer if the conclusions are wrong. It further continues until the model achieves a desired level of accuracy on the training data. This is commonly used for classification and regression. Ex. Naïve Byes, Decision tree, etc.
Used for-spam filtering, language detection, computer vision, search and classification, linear regression.
- Unsupervised learning:
No supervision, works independently. Moreover, like the process of clustering, it is often used to divide data into groups by similarity. It is good for insightful data analytics. As the human mind is unable to process large amount of numerical data, as the world of extensive business is so huge, to find fraudulent transactions, forecast sales and discounts or analyze preferences of customers. We need some patterns, and here comes the need of unsupervised learning to detect this. Ex. -DBSCAN, mean-shift, etc.
Used for-segmentation of data, anomaly detection, risk management and fake images analysis.
- Semi-supervised learning:
In this the input-data is a mixture of labelled and unlabeled samples. The programmer has in mind a desired prediction outcome but the model must find patterns to structure the data and make predictions itself.
- Reinforcement learning:
Have you ever noticed a child, who touches a hot pan, and once he does this, he never does it again, as he feels the burn and understands it is harmful? Same goes with reinforcement learning. One doesn’t need constant supervision to learn effectively, rather, by only receiving positive or negative reinforcement signals in response to one’s actions, you learn effectively. So, it is a trial and error method. It allows you to step away from training on static datasets. Instead, the computer is able to learn in dynamic, noisy environment such as game worlds or the real world,.Ex. CSGo, PubG.
Used for-self driven cars, games, robots, resource management.
Types of deep learning:
- Convolutional neural networks (CNN):
One of the most popular applied cases. Great for image/video processing or computer vision applications. These are used primarily to classify images, cluster them by similarity, and perform object recognition within scenes.
- Generative adversarial networks (GAN):
This was invented by Ian Goodfellow, who is now staff research scientist at Google Brain. A GNN is composed of two neural networks: a generative network and a discriminative network. It takes random noise signals as input and generates a random noisy(fake) image as the output. And with the help of discriminator, it starts generating images that look real.
In the early stages, when AI was founded in 1956, as an academic discipline, its aim was to achieve solutions to the logical problems. With showing successive results of AI, the researchers than, took a leap towards machine learning, that filled the gap which AI was lacking in, i.e.., how those problems are solved?
Now, machine learning, had become a professor, a guide or a trainer, you can say, that fed an algorithm a lot of data and let it figure things out. As these algorithms developed, they could tackle many problems. But some things that humans found easy9like speech or handwriting recognition) were still hard for machines. Human nature is getting ahead of what you lack behind, so if, machine learning is about mimicking how human learn, the researches further took it to next level, Deep learning, to actually mimic the human brain.
The idea of using artificial neural network was then explicitly simulated in software used for certain problems. As we know machine learning and deep learning are two major parts of AI, so we just need to know the similarities and differences to understand them better.
At a glance:
- Machine learning (ML) is the sub domain of AI and uses algorithms. Computer identify and act upon data pattern; on the other hand, Deep learning (DL) is the sub domain of machine learning. It simulates the way human brain perceives, organizes and make decisions.
- On basis of work-ML use various algorithms which help in making decisions and DL interprets data features using neural networks by passing the relevant information in the process of data processing.
- On data basis-ML has a great performance on dataset which are small to medium sized, whereas, DL performs well on large data set.
- Hardware grounds-in ML, low end machine can suffice, and in. DL high end machine with top notch configuration to assess data is needed.
- The time to execute in ML ranges from few minutes to certain hours, on the other hand, DL takes up to weeks as the process is long.
- Some algorithms can be interpreted easily in ML, and in DL it is very difficult.
As they are two different extremes but one thing, they have in common is that both are subsets of AI and both help in improving accuracy and efficiency.
Uses of AI/ML/DL
- AI/ML are driving an era of automation that holds important benefits for global industries by increasing productivity.
- They both bring in speed, to optimize internal business operations.
- They boost creativity and innovation by automating tasks so that employees focus on essential part.
- They can also handle and analyze enormous fata in an unbiased manner.
the retail sector has become hugely dependent on AI/ML.before the launch of any product, AI helps them to determine the need and the demand by analyzing the client feedback. Thus, this avoids the potential risk and losses which might occur after the release of product. Industry may also easily predict the performance of product before the launch.
a significant role of AI/ML can be seen in the financial advisory services. Most of the people seek financial advisory as to invest in what stocks they need to buy or hold or sell, here then, AI comes as a savior. It observes patterns in past data and predict how the pattern will repeat themselves in future. It also helps a person to manage their expenses by accumulating all the data from the web footprint and creating a spending graph.
AI has a lot of scope in medical science as it helps to improve diagnostic methods and in detecting diseases that might be overlooked by doctors. This helps a doctor to focus on the main part of the problem and saving them a lot of time. This process also saves a lot of money and time of the patient.
- Media and entertainment –
AI allows companies to make their content more accessible to customers. Digital media is not just an additional distribution platform but has emerged as a core revenue generator. It looks into user’s information, social media behavior and usage pattern with help of predictive analytics for better segmentation and targeting. This analysis facilitates right viewing content, and identifies the best content with maximum earning potential.
- Transportation –
AI has continuously aiming to reduce costs and generate more sales by delivering products with much shorter lead time. It detects real-time traffic detection and aid in optimizing routes thus providing faster traffic navigation.
In the long run :
As you can now evaluate there are more differences among the two then the similarities but the scientists have come too far in executing both ML/DL and giving the world a level of comfort and ease through these technologies.
As recently, we are going through a major pandemic, Corona virus, what I feel, the future totally holds on in AI. Many affluent industries, that already own AI, will further gain profits by these technologies. Our scientists and researchers will, on a running movement of ‘going local’, will further experiment and innovate new technologies to globalize India. India is the hub of brain masters, we do not lack in talent, what we lack is the economic strength. Global software vendors are after this new gold rush. The promise of new revenue has pushed software business owners to invest in AI technologies, AI will become one of the top five investment priorities for at least 30% of chief information officers. Many people, due to lack of tech-acquaintance, knowledge and skills do not trust in the capabilities of tech-solutions. But, need of protocols in this pandemic, as well as, human brain mimicking, AI/ML/DL are turning out to be integral part of life.
The thinking machine is AI’s biggest gift to mankind. The grand entry of these machines has changed the operative rules of business. Future advancements in unsupervised ML algorithms will lead to higher business outcomes. DL will be the end to end encryption.
Thank you for going through all the information I shared with you. I hope this might help you guys to decide which field you might go for. Also, it must help you clear your concepts regarding the machine era. I might not have scratched the surface of everything but I must have given you introduction to few of the topics which you always wanted to know about. The study of AI, machine learning and deep learning will come up in near future with some great achievements.
Thank you, all guys,
Would wait for your reviews