Deep Learning with PyTorch Live Course - Working with Images & Logistic Regression (Part 2 of 6)
video description
Date: 2022-03-14
Related videos
Comments and reviews: 7
Avi
Avi's Lecture Two Bookmarks (long comment)
16:40 recap of last lecture (lecture one)
27:40 instructor shows where lecture notebooks are in the forums, first one will be -03-logistic-regression-
- instructor recommends you to fork the notebook to have easier, quicker access to working with the notebook
- rather than Binder, Kaggle is better for working with larger datasets, so, instructor recommends you use Kaggle instead for working with this notebook
29:25 about Kaggle platform, that you need a free account
30:00 RECIPE (first time only) how to open & run Jovian notebook on Kaggle
- run > on kaggle
- language > python
- type > notebook
- create
- (first time only) upper right corner sidebar > internet > on
- note: once your notebook fully loads up in Kaggle, it should now have the text --live- appended onto its name
31:25 preface to lecture (comments on setup)
32:09 actual START of lecture two content with notebook -03-logistic-regression-live- and -image classification using logistic regression in pytorch-
32:21 instructor notes that we will use existing knowledge of pytorch and linear regression to solve a different problem (image classification)
32:42 about the training dataset MNIST (grayscale handwritten digit images, considered a good dataset with whihc to get started in machine learning)
33:33 instructor notes that -system setup- heading can be skipped for now since the instructor's notebook is already running
33:42 firstly, we begin by -exploring the data-
33:49 note that if you run this notebook locally, you may need to uncomment the -!conda install- or -!pip install- commands to get the necessary code libraries (these come already installed though on Kaggle)
34:07 today's lecture will include torchvision, which continues -utilities- for working with image data and -helper classes- to automatically download and import popular datasets, such as MNIST
34:34 instructor donwloads the training dataset
34:58 instructor recommends -you should read up about Yann Lecun- as -one of the forefathers of deep learning-
35:19 instructor notes that -a bunch of files were downloaded but we don't need to worry about them just yet-, the important takeaway is that we now have a -dataset- object to work with
35:32 instructor inspects the dataset object alone, and then with -len(dataset)-
36:00 instructor shows how to get out the -test dataset- from the MNIST class [constructor] (which is needed to determine the accuracy of your ML model)
36:42 instructor inspects the first element of the PyTorch dataset
37:16 instructor breaks down PyTorch dataset element shape (structure) as a PIL.Image with size dimensions and the number five. the number five here indicates that this image is showing/displaying the number five. in other words, this is a -supervised learning problem- where the training data has already be labeled (or we can simply say it contains labels - there's not necessarily a guarentee that they are perfectly correct or complete).
38:06 instructor explains that PIL stands for the the Pillow Imaging Library
38:26 we can visualize images in our notebook directly using the -cell magic- -%matplotlib inline- (which probably requires an inport)
39:17 let's inspect a couple images from the dataset
39:52 instructor notes that what is rendered in not exactly an image per se but rather a plot of pixles (colored cells)
40:40 -pytorch does not how to work with images-, so, we convert the images into Tensors using the -ToTensor()- transform
41:04 instructor shows how they convert the MNIST dataset of images and labels into tensors
reply
Avi's Lecture Two Bookmarks (long comment)
16:40 recap of last lecture (lecture one)
27:40 instructor shows where lecture notebooks are in the forums, first one will be -03-logistic-regression-
- instructor recommends you to fork the notebook to have easier, quicker access to working with the notebook
- rather than Binder, Kaggle is better for working with larger datasets, so, instructor recommends you use Kaggle instead for working with this notebook
29:25 about Kaggle platform, that you need a free account
30:00 RECIPE (first time only) how to open & run Jovian notebook on Kaggle
- run > on kaggle
- language > python
- type > notebook
- create
- (first time only) upper right corner sidebar > internet > on
- note: once your notebook fully loads up in Kaggle, it should now have the text --live- appended onto its name
31:25 preface to lecture (comments on setup)
32:09 actual START of lecture two content with notebook -03-logistic-regression-live- and -image classification using logistic regression in pytorch-
32:21 instructor notes that we will use existing knowledge of pytorch and linear regression to solve a different problem (image classification)
32:42 about the training dataset MNIST (grayscale handwritten digit images, considered a good dataset with whihc to get started in machine learning)
33:33 instructor notes that -system setup- heading can be skipped for now since the instructor's notebook is already running
33:42 firstly, we begin by -exploring the data-
33:49 note that if you run this notebook locally, you may need to uncomment the -!conda install- or -!pip install- commands to get the necessary code libraries (these come already installed though on Kaggle)
34:07 today's lecture will include torchvision, which continues -utilities- for working with image data and -helper classes- to automatically download and import popular datasets, such as MNIST
34:34 instructor donwloads the training dataset
34:58 instructor recommends -you should read up about Yann Lecun- as -one of the forefathers of deep learning-
35:19 instructor notes that -a bunch of files were downloaded but we don't need to worry about them just yet-, the important takeaway is that we now have a -dataset- object to work with
35:32 instructor inspects the dataset object alone, and then with -len(dataset)-
36:00 instructor shows how to get out the -test dataset- from the MNIST class [constructor] (which is needed to determine the accuracy of your ML model)
36:42 instructor inspects the first element of the PyTorch dataset
37:16 instructor breaks down PyTorch dataset element shape (structure) as a PIL.Image with size dimensions and the number five. the number five here indicates that this image is showing/displaying the number five. in other words, this is a -supervised learning problem- where the training data has already be labeled (or we can simply say it contains labels - there's not necessarily a guarentee that they are perfectly correct or complete).
38:06 instructor explains that PIL stands for the the Pillow Imaging Library
38:26 we can visualize images in our notebook directly using the -cell magic- -%matplotlib inline- (which probably requires an inport)
39:17 let's inspect a couple images from the dataset
39:52 instructor notes that what is rendered in not exactly an image per se but rather a plot of pixles (colored cells)
40:40 -pytorch does not how to work with images-, so, we convert the images into Tensors using the -ToTensor()- transform
41:04 instructor shows how they convert the MNIST dataset of images and labels into tensors
reply
Multitudes
1:04 I realise now that it was quite clever to have apples and oranges in the previous notebook: 2-Linear regression because I now can understand better why the shape is 10, 784. For every picture the weights are calculated ten times, one for each output case since they are the probability of the number being recognised correctly. Well done
reply
1:04 I realise now that it was quite clever to have apples and oranges in the previous notebook: 2-Linear regression because I now can understand better why the shape is 10, 784. For every picture the weights are calculated ten times, one for each output case since they are the probability of the number being recognised correctly. Well done
reply
maciej
hi, my model works really good on the Mnist data (92 % acc) but did not work on the images i fed to it at all, i draw the image, then transformed it to binary and then resize it and transform to tensor but it always predict the wrong number... what can be the case?
reply
hi, my model works really good on the Mnist data (92 % acc) but did not work on the images i fed to it at all, i draw the image, then transformed it to binary and then resize it and transform to tensor but it always predict the wrong number... what can be the case?
reply
fede
I have checked the deadlines for project 1, these deadlines are for people without other responsibilities because of this I think I am going to pass on doing this training, I have already been stressed out by trying to do many things at once. Thank you.
reply
I have checked the deadlines for project 1, these deadlines are for people without other responsibilities because of this I think I am going to pass on doing this training, I have already been stressed out by trying to do many things at once. Thank you.
reply
Talha
How we make sure data is shuffled while doing random_split, can we set shuffle=None as in keras? and if we have shuffled the dataset, why we are shuffling it again in data loader?
reply
How we make sure data is shuffled while doing random_split, can we set shuffle=None as in keras? and if we have shuffled the dataset, why we are shuffling it again in data loader?
reply
Elite
Please help me i need a one Facebook id hack this id connected free fire account and my free fire account banned so i need this fb id hack emergency pls pls pls mail me
reply
Please help me i need a one Facebook id hack this id connected free fire account and my free fire account banned so i need this fb id hack emergency pls pls pls mail me
reply
Mausam
If max function is non-differentiable and as cross-entropy has softmax calculated internally then isn't the non-differentiability inherited to cross-entropy ?
reply
If max function is non-differentiable and as cross-entropy has softmax calculated internally then isn't the non-differentiability inherited to cross-entropy ?
reply
Add a review, comment
Other channel videos
