3 days, 10 hours ago

 of course, if you input enough high-quality data to train the model, the quality of the training data will ultimately determine the prediction effect, otherwise there may be high false positive (AI predicts your girlfriend will be angry, but she does not) or false negative (AI) All of a sudden, your girlfriend didn't change her face. 


 OK, then if it was me, I would construct the AI model like this: 



 algorithm 
: deep learning


 
 
 
 group 
:

Anger group: take the anger you notice as the node, review all the data to be collected in the first 15 minutes of the node (of course, if you want to give early warning, you need to further extend the time before the gas node)

Safety and stability group: take any time when there is no gas as the node to review all the data to be collected within the first 15 minutes of the node (of course, if the acquisition time is extended, the same as above)


 parameters to be collected 
:

Immediate weight (this may be a little difficult, unless there is a scale right next to it), the type of clothes you wear on the day (Royal sister style, Lori style, JK uniform, ol Make complaints about the topic of conversation (chat, chat, chat, gossip, chat, job, Tucao...), collect the expression of time girlfriend (image data), the latest meal, bla bla bla... (feel shy, I am here to answer, I can't make it up) 


 
 
 
 data amount: 
 data size

Collect at least 100 times the data of the girlfriend angry group (
 the more data, the better, but don't deliberately provocation to prevent data distortion 
). The safe group is good at collecting data, so 200-300 groups of data are collected. 

Well, after such a set of operations, the basic model should be established. The next step is to optimize the model parameters. For example, those data are not significant. You can remove them and re model them to see if the AUC value increases (the higher the AUC value, the better the model prediction). PCA can also be used to make a classification. 


 in short, in theory, AI can be used to predict whether your girlfriend is angry. However, with such a set of coquettes, you can tell whether you have a girlfriend or not

Comment

No Comment Yet.