Comments - The AI layer for the Enterprise and the role of IoT - Data Science Central 2021-06-15T13:02:39Z https://www.datasciencecentral.com/profiles/comment/feed?attachedTo=6448529%3ABlogPost%3A472769&xn_auth=no Quote : "algorithm which is l… tag:www.datasciencecentral.com,2016-10-03:6448529:Comment:473848 2016-10-03T21:30:34.940Z Sione Palu https://www.datasciencecentral.com/profile/SioneKPalu <p>Quote : "algorithm which is learning on its own"</p> <p>No, the algorithm doesn't learn on its own. There's a huge misunderstanding in machine-learning &amp; data science community about learning. They equate the term learning as somehow equivalent to humans' learning which they are not the same.</p> <p>The learning refers to in machine learning is nothing more than the definition in mathematics &amp; statistics of:<br></br>Functional-Mapping, where input X is mapped to Y via a function 'f'. f:X…</p> <p>Quote : "algorithm which is learning on its own"</p> <p>No, the algorithm doesn't learn on its own. There's a huge misunderstanding in machine-learning &amp; data science community about learning. They equate the term learning as somehow equivalent to humans' learning which they are not the same.</p> <p>The learning refers to in machine learning is nothing more than the definition in mathematics &amp; statistics of:<br/>Functional-Mapping, where input X is mapped to Y via a function 'f'. f:X --&gt; Y</p> <p><a href="https://en.wikipedia.org/wiki/Function_" target="_blank">https://en.wikipedia.org/wiki/Function_</a>(mathematics)</p> <p>In supervised learning, we have X and Y but the job is to find the mapping function 'f'.</p> <p>This is not learning on its own. It is simply function fitting. If its a vector X and vector Y, then the mapping is called curve-fitting. The definition is that simple.</p> <p>I hope that educators of data science should educate the public by avoiding using hype language, because this is the reason I suspect that the public are sort of scary about AI and gives them a misconception.</p>