Machine Learning / Deep Learning Tutorials for Programmers playlist: ua-cam.com/play/PLZbbT5o_s2xq7LwI2y8_QtvuXZedL6tQU.html Keras Machine Learning / Deep Learning Tutorial playlist: ua-cam.com/play/PLZbbT5o_s2xrwRnXk_yCPtnqqo4_u2YGL.html
Thank you very much for this video! The use of an example and the specific explanation on freezing layers really helps me understand how fine-tuning can conceptually be done for similar tasks!
I really enjoyed your lectures. it will be grateful if you could go through an example step by step for hand-digit recognition or car or anything else. It will be much better using code from scratch rather than using prepared libraries.
I spotted a slight typo in the article for this video. recofnizing ↓ recognizing I really enjoy your courses so far, by the way. I've stopped and started a few times with studying ML in the past, but this has been a pleasure to go through.
{ "question": "Why do we freeze some of the initial layers of the network in transfer learning", "choices": [ "to use these layers without updating the weights", "to exclude these layers while training", "to update the weights of these layers", "to initialize the weights with zero for these layers " ], "answer": "to use these layers without updating the weights", "creator": "Ajeet Verma", "creationDate": "2021-05-08T09:07:14.299Z" }
Is this just a substantial amount of the If else then statement application? (Not a coder or anything, just remember some stuff from Computing at U of H) I am interested in machine learning stuff as it pertains to my theory/book I am writing. I am using some stuff from this video, that I will reach out to you about later, as I am doing the same with all others whose content I am referencing. Thank you so much for this simplistic explanation.
Thank you for your video, i have a question i have a model of speech to text in spanish (i didn't train it) i want to improve the model, can i use fine tuning right?
A quick question. Is there fine-tuning process for text classification using Keras? I see that your focus is on image classification, what about text classification?Thanks
I have a question, if I have a model that can detect license plates of cars, would it be feasible or possible even, to fine tune the model to make it able to detect license plates of motorbikes?
Hey Nilesh - Kind of. With fine-tuning, we take an existing model that has already been trained on a data set, and we "tune" it by modifying layers, parameters, etc. and re-train it on new data. So in a sense, the fine-tuned model is inheriting its parameters from the original model and then being changed incrementally to accommodate its new task/new data.
Yes, updates occur after each batch. More on this in the blog below. This kind of update is called "mini-batch gradient descent." Each epoch sees the same input data, but with batches of data shuffled in a different order. deeplizard.com/learn/video/U4WB9p6ODjM
Machine Learning / Deep Learning Tutorials for Programmers playlist: ua-cam.com/play/PLZbbT5o_s2xq7LwI2y8_QtvuXZedL6tQU.html
Keras Machine Learning / Deep Learning Tutorial playlist: ua-cam.com/play/PLZbbT5o_s2xrwRnXk_yCPtnqqo4_u2YGL.html
Brief, Clear, Understandable and Complete, without usual youtubers playing round. Thank you!
Your videos are by far the best on UA-cam for this kind of thing. Thank you for producing them.
Thank you very much for this video! The use of an example and the specific explanation on freezing layers really helps me understand how fine-tuning can conceptually be done for similar tasks!
Check out the corresponding blog and other resources for this video at: deeplizard.com/learn/video/5T-iXNNiwIs
Thanks 4 the video and good info.
Found the perfect explanation of fine tune.
All of your tutorials are amazing.Simply explained.Helps me a lot.
very help full video
Thanks for the explanation in a very simple way.
Brilliant 🙄🖐️🖐️🖐️🖐️👉💫👉🎶👉📲👉🔌👉❤️👉🙄👉👆👆👆👉🙏👉😃👉😐
What a great channel ... am surprised it still has less than 100k subs!
Thanks for simplifying the topic.
Thank you for this video!
I really enjoyed your lectures. it will be grateful if you could go through an example step by step for hand-digit recognition or car or anything else.
It will be much better using code from scratch rather than using prepared libraries.
this is so interesting...I can already see the applications for foreign language adaptation, disease classification, drug research...
also are the weights and biases updated per batch? or per epoch?
per batch
Good stuff.
great explanation
Came back to revise the concept. Still helpful.
Exactly what I needed to know before starting that finetuning tutorial ! :)
God bless you, Teacher
I spotted a slight typo in the article for this video.
recofnizing
↓
recognizing
I really enjoy your courses so far, by the way. I've stopped and started a few times with studying ML in the past, but this has been a pleasure to go through.
Fixed, thanks Chris! :D
0:10 intro
2:00 car recognition example
Great Content !! Please if you can make videos on transfer learning... Thank You.
Thank you ✨
thanks for video
best video ever
thank u, it's a very useful video
Awesome explanation as always :)
Thanks a lot !
{
"question": "Why do we freeze some of the initial layers of the network in transfer learning",
"choices": [
"to use these layers without updating the weights",
"to exclude these layers while training",
"to update the weights of these layers",
"to initialize the weights with zero for these layers "
],
"answer": "to use these layers without updating the weights",
"creator": "Ajeet Verma",
"creationDate": "2021-05-08T09:07:14.299Z"
}
amazing explanation. thanks a lot.
Thanks, Knowhow! And you're welcome :)
{
"question": "Fine Tuning related to ",
"choices": [
"Transfer learning",
"Unsupervised learning",
"Supervised learning ",
"Association learning"
],
"answer": "Transfer learning",
"creator": "Hivemid",
"creationDate": "2022-03-31T10:46:59.763Z"
}
Great job!
Thank you so much!
Thank you for the video! Can you send a link to an example how to do fine tuning?
deeplizard.com/learn/video/oDHpqu52soI
For more general neural networks like LLM, would it be possible to fine tune it to basically anything text-based?
thank you
thanks for this video
You're welcome, Lan!
you are amazing guys
Is this just a substantial amount of the If else then statement application? (Not a coder or anything, just remember some stuff from Computing at U of H) I am interested in machine learning stuff as it pertains to my theory/book I am writing. I am using some stuff from this video, that I will reach out to you about later, as I am doing the same with all others whose content I am referencing. Thank you so much for this simplistic explanation.
Thank you :)
ty
how does one freeze some weights please?
You are extraordinary :)
Thanks,
just wow
Thank you for your video, i have a question i have a model of speech to text in spanish (i didn't train it) i want to improve the model, can i use fine tuning right?
Could anyone say whether Fine Tuning and Transfer learning are same? If not, then how they are different? Thanks in advance!
A quick question. Is there fine-tuning process for text classification using Keras? I see that your focus is on image classification, what about text classification?Thanks
Can anyone say whether Fine tuning and Transfer learning are same?? Thanks in advance!
wow
I have a question, if I have a model that can detect license plates of cars, would it be feasible or possible even, to fine tune the model to make it able to detect license plates of motorbikes?
Yes
How is Fine-tuning related to Transfer Learning ?
its just like inheritance?? where we can aquires the features of parent class
Hey Nilesh - Kind of. With fine-tuning, we take an existing model that has already been trained on a data set, and we "tune" it by modifying layers, parameters, etc. and re-train it on new data. So in a sense, the fine-tuned model is inheriting its parameters from the original model and then being changed incrementally to accommodate its new task/new data.
after each batch are the weights updated?
does each epoch see the same input data?
thanks!
Yes, updates occur after each batch. More on this in the blog below. This kind of update is called "mini-batch gradient descent."
Each epoch sees the same input data, but with batches of data shuffled in a different order.
deeplizard.com/learn/video/U4WB9p6ODjM
What is steps_per_ephocs?
The number of batches that it will take to complete one epoch.
It is introduced in our blog here:
deeplizard.com/learn/video/daovGOlMbT4
How is this different from transfer learning? I figured you would be tweaking individual weights based on the title