Check this out ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-DU0aPwXXbwg.html . You will really like it. Pls leave a comment if you do, its a unlisted video :)
Check this out ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-DU0aPwXXbwg.html . You will really like it. Pls leave a comment if you do, its a unlisted video.
Quick question on 17:30 - How is the alpha argument used to build a model? Does it work on top of a trained model? Is it to be defined prior to training? How are the channels reduced? The paper is not clear at all on that. I would appreciate if you could give me some insight on this.
Hey, you said the 'N' in convolution corresponds to the number of filters but the paper says it represents the depth pf the output channels. Can you describe a little more on what you think 'N' corresponds to.
Each filter acts on the previous activation block to create a 2D activation map. If a layer has N filters then we have N such 2D maps. So if we stack them we have a depth of N.
Each filter corresponds to the depth of the output. Example if the input is HxWxC and you have 4 filters and 1x1 kernel size, the output will be HxWx4.
Ladidadee Rayo jagdeep banghra taxi jagdeep Rakesh abhinav T series t series’s taxi service would like to informing you to subscribing to pewdiepie 3 idiots