How To Train A Machine Learning Algorithm To Differentiate Right From Wrong
Humans learn from experience. Machine learning algorithms also learn from experiences, or as we like to put it—from a “machine learning corpus.”
For instance, if you want to train an algorithm to recognize dogs, you feed it myriad pictures of dogs along with images of other animals, say cows. Eventually, it can differentiate a dog from a cow. So, only after the algorithm has been fed enough data does it become useful.
However, if not done correctly, things can go awfully wrong in a matter of seconds.
Let’s cite an example of Microsoft’s Tay, a bot that the company deployed on the microblogging platform Twitter. Only a few hours into the launch, Tay started spewing bigoted tweets. It went from being friendly (“Humans are super cool”) to downright scary (“Hitler was right and I hate Jews”). Yikes!
The debacle was a classic example of a well-meaning experiment gone wrong.
With SearchUnify, such fiascos are a thing of the past. If you are a SearchUnify user for some time now, you’d already be aware of its machine-learning capabilities.
That said, aren’t you intrigued to know what goes under the hood to render relevant results to user queries?
If yes, then you’re in luck. We’re introducing an exciting feature with Mamba ‘23. To know more, you’ll have to wait a little bit longer. Rest assured, it’s worth the wait!
For instance, if you want to train an algorithm to recognize dogs, you feed it myriad pictures of dogs along with images of other animals, say cows. Eventually, it can differentiate a dog from a cow. So, only after the algorithm has been fed enough data does it become useful.
However, if not done correctly, things can go awfully wrong in a matter of seconds.
Let’s cite an example of Microsoft’s Tay, a bot that the company deployed on the microblogging platform Twitter. Only a few hours into the launch, Tay started spewing bigoted tweets. It went from being friendly (“Humans are super cool”) to downright scary (“Hitler was right and I hate Jews”). Yikes!
The debacle was a classic example of a well-meaning experiment gone wrong.
With SearchUnify, such fiascos are a thing of the past. If you are a SearchUnify user for some time now, you’d already be aware of its machine-learning capabilities.
That said, aren’t you intrigued to know what goes under the hood to render relevant results to user queries?
If yes, then you’re in luck. We’re introducing an exciting feature with Mamba ‘23. To know more, you’ll have to wait a little bit longer. Rest assured, it’s worth the wait!
0
Please sign in to leave a comment.
Comments
0 comments