Post by account_disabled on Mar 11, 2024 9:01:05 GMT
In this interview with Yann LeCun of BergamoScienza we discover how Deep Learning and AI work with some truly winning examples. Yann LeCun is one of the world's leading experts in the field of artificial intelligence. Alessio Pomaro Alessio Pomaro 21 Oct 2021 •4 min read How does artificial intelligence work? Interview with Yann LeCun How does artificial intelligence work? Interview with Yann LeCun On October 15, Yann LeCun participated in a BergamoScienza meeting entitled " New forms of intelligence ", during which he answered the most difficult questions of our times about AI. In this post I tried to summarize the most interesting concepts. Who is Yann LeCun? Yann LeCun , VP & Chief AI Scientist at Meta and professor at New York University, is one of the world's leading experts in the field of Artificial Intelligence .
Yann LeCun, VP & Chief AI Scientist at Meta and professor India Mobile Number Data at NYU Yann LeCun, VP & Chief AI Scientist at Meta and professor at NYU What is Deep Learning and how is it different from Machine Learning? Why did it revolutionize artificial intelligence? Let's start by understanding what machine learning is . If we want a machine to do something, we must program it by giving it instructions to carry out certain operations. In the case of machine learning, you train the machine to learn! BergamoScienza: interview with Yann LeCun For example, if you want to train a machine to translate from English to Italian , you use already translated sentences to explain to the system the result you want to obtain. If, however, you want to train to recognize an object within images , for example a cat, you show the image of the cat to the machine, and if it doesn't recognize it you correct it, telling it that it is indeed a cat.
At that point the system adapts, corrects itself automatically and constantly. So, the next time, he will be able to recognize the subject. The one described is supervised machine learning . These techniques have been developed over the last 50 years. Deep Learning has to do with the construction of an artificial neural network made up of many elements connected to each other. The connections and the strength of these connections represent values that can be corrected to make the machine, for example, recognize an element in the image. When the system doesn't recognize it, we adjust the connections ( perhaps even hundreds of them in the neural network ) to ensure that the response comes as close as possible to what we expected.
Yann LeCun, VP & Chief AI Scientist at Meta and professor India Mobile Number Data at NYU Yann LeCun, VP & Chief AI Scientist at Meta and professor at NYU What is Deep Learning and how is it different from Machine Learning? Why did it revolutionize artificial intelligence? Let's start by understanding what machine learning is . If we want a machine to do something, we must program it by giving it instructions to carry out certain operations. In the case of machine learning, you train the machine to learn! BergamoScienza: interview with Yann LeCun For example, if you want to train a machine to translate from English to Italian , you use already translated sentences to explain to the system the result you want to obtain. If, however, you want to train to recognize an object within images , for example a cat, you show the image of the cat to the machine, and if it doesn't recognize it you correct it, telling it that it is indeed a cat.
At that point the system adapts, corrects itself automatically and constantly. So, the next time, he will be able to recognize the subject. The one described is supervised machine learning . These techniques have been developed over the last 50 years. Deep Learning has to do with the construction of an artificial neural network made up of many elements connected to each other. The connections and the strength of these connections represent values that can be corrected to make the machine, for example, recognize an element in the image. When the system doesn't recognize it, we adjust the connections ( perhaps even hundreds of them in the neural network ) to ensure that the response comes as close as possible to what we expected.