Now you can have more fluid conversations with Alexa with its new machine learning function
A movie recently came out about an average guy - friendless and addicted to the phone - who has a strange relationship with his personal assistant on his smartphone, and they even have sex in some strange way. The movie is titled Jexi, and it seems that fiction has caught up with us (again). According to Amazon, interacting with the latest model of your virtual assistant Alexa will be like talking to another human being.
Through a statement, Jeff Bezos' company, the Lex Luthor of technology, reported that the new version of Alexa will include a new machine learning system that aims to make “customers find interacting with it (Alexa) as natural as interacting with another human being, ”they said. In theory, the new Alexa feature will allow you to perceive the "latent targets" of user requests. In other words, it will anticipate customer requests.
“The latent target discovery model analyzes multiple characteristics of customer statements, including mutual point information, which measures the likelihood of an interaction pattern in a given context relative to its likelihood across all Alexa traffic. Deep learning-based sub-modules evaluate additional characteristics, such as whether the client was trying to rephrase a previous command or issue a new command, or whether the direct target and the latent target share common entities or values (such as required time value to steep the tea), ”Amazon explained.
If a user asked Alexa how long it might take to make a cup of coffee, the assistant could not only answer the estimated brewing time, but could also suggest setting a timer ... without the user having previously requested it.
According to Amazon, the new Alexa function is based on a series of sophisticated algorithms to anticipate the requests of its users and a deep learning model, which will evolve based on its relationship with the user. If every morning when waking up the user asked Alexa about the preparation of coffee and requested the timer, over time it would anticipate the situation.
Date update on 2020-11-15. Date published on 2020-11-15. Category: Computer class Author: Oscar olg Fuente: codigoespagueti