I don't understand somthing in week 2 convolution deep leaning specialization

#Deep learning specialization
#Convolution
#Week 2
#Last video(state of deep computer vision)

I dont undrastand what does “carfully design feature” exacly mean and it use “simple algoritm” and “gient neural network” together so i have confused
Can someone help me : this is a text of course
" if you look across a broad spectrum of machine learning problems, you see on average that when you have a lot of data you tend to find people getting away with using simpler algorithms as well as less hand-engineering. So, there’s just less needing to carefully design features for the problem, but instead you can have a giant neural network, even a simpler architecture, and have a neural network. Just learn whether we want to learn we have a lot of data. Whereas, in contrast when you don’t have that much data then on average you see people engaging in more hand-engineering"

When we have a lot of data, people don’t spend too much time in manually engineering features. They throw the problem on the model and it learns well.
When data isn’t available in plenty, best results are often achieved when manual feature engineering is performed.

1 Like

Does Engineering feature mean use shift color and crop and …?

Shifting/translating/color/crop are examples of Data Augmentation.
Data Augmentation is one way of Feature Engineering.

So yes, shifting/translating/color/crop are part of Feature Engineering.

Feature Engineering is a broad term that could mean many other things as well, such as selecting the right features, normalizing, removing bad/incorrect/missing data, etc.

1 Like

Ty but what about my second question what does simpler algoritm mean here ? He said simpler algoritm and giant neural network when we have a lot of data . So,what does algoritm here ? I get confused because i think giant network make our algorithms difficult

Hello @mhaydari81,

If we have hand-engineered sufficient features that predict the labels well, we can just send them to a simple logistic regression algorithm. Below is an example of how neural network may replace hand-engineering.

Cheers,
Raymond