- The Generations of AI
- Standard Definition of AI & ML
- AI ML Applications in Real world
- The Evolution of AI ML in the Industry
- AI through Infographics
- The world of Deep Learning applications
- The world of AI Coding is also waiting to welcome you with its warmth and plethora of actual code sets from REAL WORLD of AI applications.
- Please keep checking your compass for seamlessly steer your way to the next mile of this wonderful journey.
The Generations of AI
The 1st Generation AI: The first generation of AI was rules-based and emulated classical logic to draw reasoned conclusions within a specific, narrowly defined problem domain. It was well suited to monitoring processes and improving efficiency, for example.
The 2nd Generation AI: The second, current generation is largely concerned with sensing and perception, such as using deep-learning networks to analyze the contents of a video frame.
The 3rd Generation AI: A coming next generation will extend AI into areas that correspond to human cognition, such as interpretation and autonomous adaptation. Next-generation AI must be able to address novel situations and abstraction to automate ordinary human activities.
- Q) What Is New in 3rd Gen AI? - A) This will overcome the so-called “brittleness” of AI solutions based on neural network training and inference. - Q) What do you mean by the so-called “brittleness” of AI solutions? - A) Deep neural network training and inference heavily depend on literal, deterministic views of events that lack context and commonsense understanding.
Standard Definition of AI and ML
Simply put across, AI is described as as any task performed by a program or a machine that requires application of human like intelligence to accomplish the task. Emulate human cognition | Simulation of humane intelligence . It 's technical simulation.i.e,- tecnology which uses complex algorithmic techniques to simulate the way neurons works in human brain. Neurons are the basic unit of our nervous system.
AI is superset of Machine learning, Cognitive learning and deep learning, Reinforcement Learning.
ML is algorithmic & satistical approach to approximate conclusions, predictions without direct human input. Without being Explicitly programmed. Does Patten finding. Data driven decisions.
AI ML Applications in Real world
This section offers just a sneak peek for couple of domains/sectors for to help you get the intuition into AI ML Applications in Real world. The detailed curated list is present in the main repository.
Sneak peek into AI in Healthcare
Most of the medical and healthcare-related advancements are powered by AI such as
- Enhanced the precision of robot-assisted surgery
- Clinical research and developments - such as Clinical trials
- Automated analysis of cardiac MRI scans/ CT scan/ MRI scans /X rays for assisting radiologists
- Automated, Sensitive,, Cost-effective Intervention and personalize treatment
Interesting tidbit -
- AI is better than many dermatologists at diagnosing skin cancer.
- In a study published in the leading cancer journal - Annals of Oncology, Y2018 -
- Dermatologists were only 86.6% accurate at diagnosing skin cancer,
- while the computer was able to diagnose issues with a 95% accuracy. It was also quoted in Fortune magazine published in Y2018.
Sneak peek intoAI in Finance
- Risk Assessment
- Fraud Detection And Management
- Financial Advisory Services
Advanced application Brain Models and Simulation
Can you imagine a brain and its workings being replicated on a computer? That is what the Brain Simulation Platform (BSP) aims to do. The BSP is available to researchers worldwide, so that they can compare their experimental results with model predictions and conduct investigations that are not possible experimentally.
Simulation also aims to replicate work on animal models, such as the mouse. In addition, the computing environment used for simulation offers the possibility of studying disease processes electronically.
However, the challenge is a complex one, as the human brain contains 86 billion brain cells (known as neurons) each with an average of 7,000 connections to other neurons (known as synapses). Current computer power is insufficient to model a entire human brain at this level of interconnectedness. Quantum Computing is now assisting BSP.
The Evolution of AI ML in the Industry
Well, now Letz get cracking on the evolution of AI over the time..!!
First thing first, The field of AI is not new, rather it dates back to 1950s. Infact, Artificial intelligence was founded as an academic discipline in 1955. & since then, AI has experienced several waves of improvement, marked by new approaches and new advancements. It is also worth noting that the field of AI had its fair share of waves of disappointments - which is called as AI winter. & AI winter phses were marked by dismal academic interest and loss of industrial funding in AI. But, today AI is ubiquitous. Today AI is no more novelty, rather it is a necessity. We all very well know It's pretty much part of our day today life.
so Lets have a look at some of the milestones in AI research & development.
• Turbochamp in 1950. Alan Turing Created a Chess Computer Program, called Turbochamp. Alan Turing who is widely considered to be the father of AI. back then, Turbochamp logic was based on just a few of the most basic rules of chess, and it was only able to “think” two moves in advance.
A fun fact & Just a quick reminder - Fast forward to 1997, Deep Blue (a chess-playing system developed by IBM) could beat Garry Kasparov who was then a world chess champion.But, nevertheless, Turbochamp certainly marks the beginning of evolution of AI.
• In the late 1960s, computer scientists started working on Machine Vision Learning which is now called as Computer vision- a specialized interdisciplinary division of AI. and leading Universities across the globe started developing machine learning in robots. One of such leading initiative came from Waseda University, which is based out of Tokyo in Japan. This university initiated the WABOT project. WABOT - stands for WasedaRobots - which is conceptualized as Humanoid robot. (Humanoid - human like). Infact, WABOT project marks the begining of histrory of Humanoids.
• Waseda university successfully developed WABOT-1 project In the early 1970s. WABOT-1 was the world's first full-scale general purpose humanoid intelligent robot. It was the first humnaoid, able to walk, communicate to certain extent. It can use it's vision system to measure distances and directions to the objects. and grip and transport objects with hands. Note please : It was a general purpose humanoid.
• The advancements in this area of research continued in 1980s. In 1980s, the same waseda university came up with successful WABOT-2 project. & this time, WABOT-2 was not a general purpose robot like the WABOT-1. It was was rather a specialist robot. & why so? Coz WABOT-2 was a musician humanoid robot able to
- communicate with a person to certain extent, - read a normal musical score with its vision and - play keyboard instruments.
Artistic activity such as playing a keyboard instrument does require human-like intelligence. therefore, WABOT-2 marks great advancement in the field of AI, to be specific - machines with human like intelligence. This was indeed a major milestone in the field of AI and ML.
• Then came A.L.I.C.E. in mid 1990s. also referred to as Alicebot. with a full form which is Artificial Linguistic Internet Computer Entity. IT is a natural language processing chatterbot —a program that engages in a conversation with a human. It was powered by heuristical pattern matching rules techniques.
This to me is a major milestone in the field of NLP & NLU which is specific interdisciplinary sub-domain of AI and ML.
• In the late 90s and early 21st century, AI began to be used for data mining, medical diagnosis and other areas. Infact I would label Y2000-to-Y2012 as the phase of deep learning or deep neural networks which has revolutionized the field of AI. Deep Neural networks is the technique which has made AI part of our day today life. virtual assistants such as Amazon's Alexa Apple's Siri, or Google Assistant, the recommendation engine which tells you what you should buy next online, automated face recognition, object detection, or automated detection of credit card fraud and many more like this are powered by this techniques.
- Geff Hinton together with
- Yan le cunn, and
- Yoshua Bengio trigged the success of Deep neural networks.
Infact Geff hinton together with Yan le cunn, Yoshua bengio are regraded as Godfather of Deep Learning.
Well, an interesting tidbit for you all -
- Yann le cunn heads the Facebook AI Research labs.
- Geff Hinton is working for Google.
AI through Infographics
"A picture is worth a thousand words." // "One look is worth a thousand words." _Frederick R. Barnard
Therefore, my endeavor is to introduce you the wonderland of AI through the info-graphics! Please relish the wonderland!
The world of Deep Learning applications
The world of Deep Learning application with latest architectures and models! Deep Learning applications, leverage latest architectures and models, can be built for following technical domains -
- Computer Visions (CV) - Image & Video Analytics
- NLP/ NLU/ NLG - Natural Language Processing
- Audio & Speech Analytics
- Recommendation Engine
- MultiModal service (fusion of 2 domains, such as Im2Txt -Image Captioning, Q&A from Image, et al)
Predictive analytics on structured data
- Time series forecasting
- GAN (such as for fake face generation)
- Text to Image generation