Machines capable of learning, reasoning, self correction are known as intelligent machines having artificial intelligence. In short machine intelligence is Artificial Intelligence. Artificial Intelligence is achieved through hardware that resembles human brain as in neuromorphic computing. IBM has developed a brain inspired architecture using neurosynaptic cores. Most machines today are based on Von Neumann architecture. To achieve machine intelligence researchers are combining Von Neumann architecture with neuromorphic architecture. The Von Neumann architecture activity is measured in Flops while neuromorphic architecture activity is measured in sops or synaptic operations per second. Neuromorphic architecture simulates the working of neurons using neural networks. This helps researchers achieve amazingly high computing ability than traditional computers at lesser energy consumption as demonstrated by IBM true north board and Intel's Loihi board. Memristors also are used to achieve cognitive computing in machines or machine intelligence.
A concise history of AI
Concept of artificial intelligence has existed for 1000s of years. The term 'Artificial Intelligence' became official at 1956 Dartmouth Summer Research Project. John McCarthy and Marvin Lee Minsky are the Founding fathers of modern artificial intelligence. The Dartmouth conference also outlined 7 original aspects of artificial intelligence. They are-
1) Automatic computers
2)How computers can be programmed to use a language
3)Neural nets
4)Ability to measure the complexity of a problem
5) Abstraction
6) Self improvement
7) Randomness and creativity
Jeff Hinton is yet another prolific researcher in this field who kept working on developing AI through thick and thin. Research on artificial intelligence boomed from 1956 to 1974 with lots of funding and researchers taking interest. The period from 1974 to 1980 is known as first AI winter. Money stopped flowing and researcher interest waned due to lack of computing power and data. Period from 1980 to 1987 is known as second AI boom. During this period expert systems rejuvenated researcher interest. Hopfield net was created. Period from 1987 to 1993 is known as second AI winter. From 1993 onwards AI research and development exploded. Now a days we have deep neural nets, natural language processing and facial recognition working seamlessly.
Artificial Intelligence, Machine Learning and Deep Learning
A machine is called intelligent if it shows cognitive capabilities such as learning and problem solving. AI can be subdivided into narrow AI, general AI and strong AI. When the machine is better than humans in a specific task, it is called narrow AI. When the machine shows human like intellectual capabilities, its general AI. When the machine is better than humans in intellectual tasks, its strong AI.
Machine Learning is a subset of Artificial Intelligence. It is the ability of a computer to analyze a large set of data and learn from it . Using machine learning algorithms, computers can learn to accomplish a task without being programmed for it.
Deep Learning is subset of machine learning. It teaches computers to learn by example. Deep Learning is how driverless cars recognize a stop sign or other cars on road. Data is fed into neural networks in order to learn the characteristics of an object and thus identify it.
Supervised Learning, Unsupervised Learning and Reinforcement Learning
Supervised Learning, Unsupervised Learning and Reinforcement Learning are types of machine learning. When the machine learns under guidance using labeled data, it is called supervised learning. When the machine learns without any guidance using unlabeled data, it is called unsupervised learning. Reinforcement learning is a type of machine learning in which an agent interacts with its environment in order to maximize the chance of a reward. Regression and classification problems can be solved using supervised learning. Association and clustering problems can be solved using unsupervised learning. In case of reinforcement learning the input depends on action taken, there is no predefined data. Supervised learning can be used to forecast sales. Unsupervised learning can be used for anomaly detection. Reinforcement learning is used to make driverless cars and gaming agents.
What is Automated Machine Learning or Auto ML
If machine learning is applied to real world problems automatically, it is identified as automated machine learning. Using traditional machine learning methods to solve real world problems is time consuming and challenging. Auto ml makes machine learning simple to use through sophisticated algorithms capable of fetching most relevant information from data set. Not everything is yet automated though. Some human expert may still be required for a successful deployment and running.
What is AI bias?
AI systems are only as good as the data they are trained on. Bad data may result into a biased AI. Algorithms created by biased individuals may show bias when used to solve problems. Bias can enter into AI system at many stages such as understanding the problem, collecting data and preparing data. More than 180 human biases have been identified. Biases can cause trust deficit between man and machine. Good thing is it can be tackled using specific methods. Identifying and solving AI bias is essential.
Future of AI and applications
AI is predicted to write a high school essay by 2026. By 2028 it will be able to generate a creative video. By 2049 it will be able to write a new york times best seller book. By 2050 it will become able to compete in a standard math competition and by 2059 it will be smart enough to conduct math research. List of AI applications is huge. AI is finding applications in transport industry, manufacturing and agriculture. It is also used to create new tech experiences. Offices worldwide are employing AI solutions for higher productivity. It is estimated to add $16 trillion to global economy by 2030. Its predicted to create more jobs than number of jobs it will eradicate. Future seems bright for artificial intelligence.
Reference:
1)https://www.britannica.com/technology/artificial-intelligence
2)https://futureoflife.org/background/benefits-risks-of-artificial-intelligence/?cn-reloaded=1
3)https://www.tutorialspoint.com/artificial_intelligence/artificial_intelligence_overview.htm
A concise history of AI
Concept of artificial intelligence has existed for 1000s of years. The term 'Artificial Intelligence' became official at 1956 Dartmouth Summer Research Project. John McCarthy and Marvin Lee Minsky are the Founding fathers of modern artificial intelligence. The Dartmouth conference also outlined 7 original aspects of artificial intelligence. They are-
1) Automatic computers
2)How computers can be programmed to use a language
3)Neural nets
4)Ability to measure the complexity of a problem
5) Abstraction
6) Self improvement
7) Randomness and creativity
Jeff Hinton is yet another prolific researcher in this field who kept working on developing AI through thick and thin. Research on artificial intelligence boomed from 1956 to 1974 with lots of funding and researchers taking interest. The period from 1974 to 1980 is known as first AI winter. Money stopped flowing and researcher interest waned due to lack of computing power and data. Period from 1980 to 1987 is known as second AI boom. During this period expert systems rejuvenated researcher interest. Hopfield net was created. Period from 1987 to 1993 is known as second AI winter. From 1993 onwards AI research and development exploded. Now a days we have deep neural nets, natural language processing and facial recognition working seamlessly.
Artificial Intelligence, Machine Learning and Deep Learning
A machine is called intelligent if it shows cognitive capabilities such as learning and problem solving. AI can be subdivided into narrow AI, general AI and strong AI. When the machine is better than humans in a specific task, it is called narrow AI. When the machine shows human like intellectual capabilities, its general AI. When the machine is better than humans in intellectual tasks, its strong AI.
Machine Learning is a subset of Artificial Intelligence. It is the ability of a computer to analyze a large set of data and learn from it . Using machine learning algorithms, computers can learn to accomplish a task without being programmed for it.
Deep Learning is subset of machine learning. It teaches computers to learn by example. Deep Learning is how driverless cars recognize a stop sign or other cars on road. Data is fed into neural networks in order to learn the characteristics of an object and thus identify it.
Supervised Learning, Unsupervised Learning and Reinforcement Learning
Supervised Learning, Unsupervised Learning and Reinforcement Learning are types of machine learning. When the machine learns under guidance using labeled data, it is called supervised learning. When the machine learns without any guidance using unlabeled data, it is called unsupervised learning. Reinforcement learning is a type of machine learning in which an agent interacts with its environment in order to maximize the chance of a reward. Regression and classification problems can be solved using supervised learning. Association and clustering problems can be solved using unsupervised learning. In case of reinforcement learning the input depends on action taken, there is no predefined data. Supervised learning can be used to forecast sales. Unsupervised learning can be used for anomaly detection. Reinforcement learning is used to make driverless cars and gaming agents.
What is Automated Machine Learning or Auto ML
If machine learning is applied to real world problems automatically, it is identified as automated machine learning. Using traditional machine learning methods to solve real world problems is time consuming and challenging. Auto ml makes machine learning simple to use through sophisticated algorithms capable of fetching most relevant information from data set. Not everything is yet automated though. Some human expert may still be required for a successful deployment and running.
What is AI bias?
AI systems are only as good as the data they are trained on. Bad data may result into a biased AI. Algorithms created by biased individuals may show bias when used to solve problems. Bias can enter into AI system at many stages such as understanding the problem, collecting data and preparing data. More than 180 human biases have been identified. Biases can cause trust deficit between man and machine. Good thing is it can be tackled using specific methods. Identifying and solving AI bias is essential.
Future of AI and applications
AI is predicted to write a high school essay by 2026. By 2028 it will be able to generate a creative video. By 2049 it will be able to write a new york times best seller book. By 2050 it will become able to compete in a standard math competition and by 2059 it will be smart enough to conduct math research. List of AI applications is huge. AI is finding applications in transport industry, manufacturing and agriculture. It is also used to create new tech experiences. Offices worldwide are employing AI solutions for higher productivity. It is estimated to add $16 trillion to global economy by 2030. Its predicted to create more jobs than number of jobs it will eradicate. Future seems bright for artificial intelligence.
Reference:
1)https://www.britannica.com/technology/artificial-intelligence
2)https://futureoflife.org/background/benefits-risks-of-artificial-intelligence/?cn-reloaded=1
3)https://www.tutorialspoint.com/artificial_intelligence/artificial_intelligence_overview.htm
4 comments:
Useful information
Keep posting
But how safe is your physical money going to be in case of a natural disaster, or a wildfire that burned down your home, or an earthquake? What about a hurricane with a huge tidal surge, a tsunami, or a major river which jumped its banks machine learning course hyderabad
Start Data Science online course today with 360DigiTMg and be ready when the job opportunity presents itself.
Data Science in Bangalore
Start your Data Science Course with 360DigiTMG and pave the way to your successful career with a job-ready curriculum and real-time projects that will aid you in bagging a high-paid job.
Data Science Training in Jodhpur
Post a Comment