What is Transfer Learning? Transfer Learning in Keras | Fine Tuning Vs Feature Extraction

Поділитися
Вставка
  • Опубліковано 31 гру 2024

КОМЕНТАРІ • 157

  • @Jannatkhan-vz1nj
    @Jannatkhan-vz1nj 2 роки тому +126

    I can't enroll in deep learning paid courses due to my weak financial conditions but due to Xcampus I am able to learn deep learning :') This channel is a blessing for peoples like me

  • @harshitdubey8673
    @harshitdubey8673 2 місяці тому +2

    since 2 month I was trying to find out the difference between finetuning and feature extraction but there were no such video that can explain this 2 topic in this easy manner, thanks to you Nitish

  • @himanshumangoli6708
    @himanshumangoli6708 2 роки тому +15

    आपकी तारीफ मे
    मै इतना समझदार और होशियार हू तोह मे क्या करू youtube channel खोल के मुफ्त मे ज्ञान फैलाउ
    Thanks brother

  • @kushagra4401
    @kushagra4401 2 роки тому +34

    this channel is just gold for machine learning enthusiasts

  • @suryashekhawat6965
    @suryashekhawat6965 3 місяці тому +2

    Mark my word! This channel will be the best channel for data science community. No PR just pure content. People will flow lessely follow this channel just need to here once. Keep up the good work man.

  • @hammadfaheem5979
    @hammadfaheem5979 2 роки тому +5

    I already learned these concept but jo clarification aap ki video se aati he wo ab kisi english video se nhi. The Best Machine Learning channel in Hindi/Urdu. Love from Pakistan.

  • @lakshminarayanakaveti856
    @lakshminarayanakaveti856 9 місяців тому +2

    I think campus x needs more subscribes and his explanation was great.no other channels can not explain in this way....

  • @mdriad4521
    @mdriad4521 Рік тому +2

    This lecture deserves 1 million views.

  • @yibenthung
    @yibenthung Рік тому +2

    I just love the way you simplified the concepts like transfer learning, fine tuning, and the architecture of the CNNs. Thank you bhaiya🙏

  • @ankanmazumdar5000
    @ankanmazumdar5000 2 роки тому +1

    The most amazing thing is I come to watch your videos to get the cream layer of the knowledge

  • @SaiKiranAdusumilli
    @SaiKiranAdusumilli Рік тому +4

    just amazing bro... ❤️.. even 10th class student can understand your explanation... straight simple clear

  • @rhack771
    @rhack771 2 місяці тому

    I am fortunate to have found your channel. Phenomenal work that you have been doing in democratising ML knowledge. Thank you Nitish!

  • @tarun94060sharma
    @tarun94060sharma Рік тому +3

    No other channels explained this topic better than your explanation.

  • @PoojaYadav-tq3yp
    @PoojaYadav-tq3yp Рік тому +1

    Your teaching style and concepts are tremendous. Thank you sir.

  • @stevesamson-p2k
    @stevesamson-p2k Рік тому +1

    Very Good One of the best videos on deep learning

  • @NasimaaAkhtar
    @NasimaaAkhtar Рік тому +1

    best video on UA-cam channel thanks for sharing

  • @ashishmalhotra2230
    @ashishmalhotra2230 Рік тому +1

    The hard work you put into your videos is amazing.

  • @pythoninhindi
    @pythoninhindi 2 місяці тому

    00:02 Transfer learning is crucial due to data and time constraints in training deep learning models.
    03:14 Transfer learning allows for using pre-trained models in your projects.
    06:40 Transfer learning enables leveraging pre-trained models for new tasks
    09:21 Transfer learning applies knowledge from one domain to another
    12:26 Transfer learning involves using pre-trained model knowledge in a new task
    15:49 Transfer learning involves extracting features and fine-tuning for specific tasks.
    19:03 Transfer learning involves fine tuning the last layers.
    22:28 Implementing Transfer Learning in Keras
    26:03 Reducing overfitting through data augmentation
    28:48 Transfer learning in Keras involves fine-tuning and feature extraction.
    31:15 Transfer learning involves fine tuning and feature extraction
    33:40 Transfer learning can greatly enhance learning
    Crafted by Merlin AI.

  • @rb4754
    @rb4754 6 місяців тому +1

    Thank you Nitish for "Transferring your Learning" to all the students and learners who wish to be a deep learning expert....

  • @shashwatManglamJain
    @shashwatManglamJain 8 місяців тому +2

    SIR HEAD OFF TO YOU THIS WAY YOU UNDERSTAND CONCEPT SO EASILY DIE HEART FAN OF YOU

  • @shashankbarai1809
    @shashankbarai1809 21 день тому

    Bhai aaap ko bohot saari doooaaye, kya tha main , kya ban gya aapse sikh ke. god bless you :)

  • @rahulagowda
    @rahulagowda 11 місяців тому +1

    the way u teach is just fab sir

  • @maddybhatt5552
    @maddybhatt5552 Рік тому +1

    one of the best videos of the list...Campusx u rock

  • @digilectures1239
    @digilectures1239 2 роки тому +1

    Due to your best way of teaching, I subscribed your channel and I hope you will upload more useful and knowledgeable videos.

  • @ashishprasadverma9428
    @ashishprasadverma9428 2 роки тому +4

    Thankyou very much sir,I don't have word to say thank you,unbelivable lecture sir.❤❤❤❤❤

  • @aniketjoshi7817
    @aniketjoshi7817 2 роки тому

    I pressed the like button before watching the video because I knew, it would be the best one.

  • @varutriparihar981
    @varutriparihar981 5 місяців тому

    no matter how much thankyou i say it will still be less. really sir thankyou so much.

  • @divyanshubse
    @divyanshubse 10 місяців тому +1

    You are great teacher sir.❤❤
    College Wale to free ki fess le rahe hai

  • @bharatreddy6728
    @bharatreddy6728 Рік тому +1

    Really Loved the entire explanation about Transfer Learning!!🔥

  • @gurjaspartap
    @gurjaspartap Рік тому +1

    Amazing explanation sir
    thanks a lot for provinding free content

  • @abhikantbhave
    @abhikantbhave Рік тому +4

    Thanks Nitish. this video is superb.
    if anyone is getting the error , import these libraries. (keras library is changed for load_img and img_to_array
    from keras.utils import img_to_array, load_img
    from keras.applications.vgg16 import preprocess_input
    from numpy import expand_dims
    also use "keras.Model" instead of only "Model" to get the layers,
    use this --> model = keras.Model (inputs=model.inputs,outputs=model.layers[1].output)
    instead of -->model = Model (inputs=model.inputs,outputs=model.layers[1].output)

  • @sameeranzingre9225
    @sameeranzingre9225 9 місяців тому

    Nice video! Btw your mic is so good I got bluffed twice by the background noise.

  • @thelife5628
    @thelife5628 Рік тому

    Maza aa gya sir...!! Transfer Learning is Over Powered OP🔥🔥🔥🔥🔥🔥!!!

  • @shitizgoel5027
    @shitizgoel5027 9 місяців тому

    Really very nice video (Both explanation and coding part ) Sir!
    Thank you very much!

  • @SurjeetSingh01
    @SurjeetSingh01 Рік тому

    wow, explained all the things in very easy manner !!

  • @ajitnayak2919
    @ajitnayak2919 2 роки тому

    Thankyou very much sir, amazing and total understandable lecture sir. 🙏🙏🙏🙏🙏🙏🙏🙏🙏🙏🙏

  • @tantranathjha3397
    @tantranathjha3397 10 місяців тому

    Great no words for such extraordinary skill

  • @ShashikantPrabhakar
    @ShashikantPrabhakar 2 роки тому

    thank you so much sir, you have cleared my all doubts related to transfer learning..

  • @abhishekvashistha2398
    @abhishekvashistha2398 Рік тому +1

    I have a question. When u applied the data augmentation why did not no of images in train and test folder increase.? the no of files were same before and after the data augmentation part.

  • @hassanjaved4730
    @hassanjaved4730 Рік тому

    absolutely brilliant even though I have the detailed concept of the CNN all families alogs but still this video help to understand the coding structure of it it is really ammazing

  • @waseemrandhawa5658
    @waseemrandhawa5658 2 роки тому

    sir aap ko dekh k feeling blessed 🙂

  • @divyanshubse
    @divyanshubse Рік тому +1

    Sir please make video on how to become a data scientist or complete roadmap to become a data scientist

  • @deeppatel_
    @deeppatel_ Рік тому

    thank you for too much information which I wanted to gain ❤❤

  • @siliconvision
    @siliconvision 2 роки тому

    best tutorial ever on transfer learning....😍😍

  • @nightowl5136
    @nightowl5136 9 місяців тому

    You explained so well..👍

  • @ankitgaidhar3324
    @ankitgaidhar3324 9 місяців тому

    ThankYou so much sir for amazing Explanation.

  • @fatamajohora5497
    @fatamajohora5497 Рік тому

    Love from Bangladesh ❤️

  • @abdulqadar9580
    @abdulqadar9580 Рік тому

    Great and honest teaching

  • @sanjeevgupta8435
    @sanjeevgupta8435 Рік тому

    This is really awesome .. Aisey hi hum kya deeplabv3 aur mobilenetv3 bhi pretrained models use kar saktey hain kyaa..?

  • @dheerajsinghpatel5144
    @dheerajsinghpatel5144 2 місяці тому +1

    Hi NItish, Can you also cover RCNN, Fast-RCNN, Faster-RCNN & YOLO Algorithm in CNN.
    It would be of great help. Thanks!!

  • @fahadriaz1357
    @fahadriaz1357 8 місяців тому +1

    Sir great explanation!

  • @AICC2222
    @AICC2222 7 місяців тому

    Very good explanation sir.

  • @sk.tuhinislam4253
    @sk.tuhinislam4253 Рік тому

    Hi, I like ur videos so much as these are very effective to learn the foundation of any topic.Tnx for the great efforts.
    I have a question.
    If image net data has already dog and cat class, why you retrained again? If I want to find similar objects like mobile which looks similar based on a object input like any mobile but these type object is not included in image net data set I mean mobile then what is the best way to use teansfer learning to get the job done?

  • @abhaykumaramanofficial
    @abhaykumaramanofficial 2 роки тому +3

    Please start mentorship program.. And also guide me for PhD in NLP domain..

  • @armaan6856
    @armaan6856 Рік тому +1

    In the 3rd notebook im getting some problem
    The accuracy is just stuck around 50% for some reason

  • @palashlalani6785
    @palashlalani6785 10 місяців тому

    @campusX where are the label file in the cats vs dog dataset I can see only Image files in it.

  • @arkanag3115
    @arkanag3115 2 роки тому +4

    WHAT ARE THE PLANS FOR RESTARTING THE MENTORSHIP PROGRAM?

  • @darshanayenkar
    @darshanayenkar 2 роки тому

    Sir plz make a video on LSTM, your video helps me a lot .... thank you

  • @ujwalbhattarai2953
    @ujwalbhattarai2953 8 місяців тому

    how do i use this for multiclass classification , tried it but got lower accuracy

  • @narendraparmar1631
    @narendraparmar1631 10 місяців тому

    Nicely Explained , Thanks

  • @_SEJALLOYA
    @_SEJALLOYA Рік тому

    Tried the fine tuning code for vgg16 pretrained model on multiclass dataset.
    #no. of wrong predictions
    wrong =0
    for i in range(len(y_true)):
    if(y_true[i] != y_pred[i]):
    wrong = wrong+1
    print("total wrong predictions: ",wrong)
    print("total correct predictions: ", (len(y_true)-wrong))
    Gave output as: total wrong predictions: 1843
    total correct predictions: 38
    Although the validation accuracy is nearly 96%, but still the predictions are wrong.
    Any reason why???

  • @ShivamPatel-jm2iy
    @ShivamPatel-jm2iy 7 місяців тому

    What is the difference between Transfer learning and domain adaptation?

  • @krishyadav6993
    @krishyadav6993 2 роки тому +1

    What is difference between Continual Learning and Transfer Learning?
    Also, What if I Want to implement continual or Transfer learning but I don't have pre-trained model like u mentioned VGG16,Restnet (So, Dataset might me in this case would be in tabular form e.g CSV data)
    Do you have any notes or code for continual learning on Tabular data?

  • @prajwaldeepkhokhar1905
    @prajwaldeepkhokhar1905 Рік тому

    How many of the layers we should unfreeze in fine tuning?

  • @pawannawade
    @pawannawade 2 роки тому

    I have one question ❓
    Which books you are preferred to get this deep knowledge of each and every topic.
    Please reply me sir with your valuable answer.

  • @yashmatlani1976
    @yashmatlani1976 5 місяців тому

    Please make vidoes on transfer learning methods for NLP

  • @aninditasaha0307
    @aninditasaha0307 6 місяців тому

    Can you apply this on UCIHAR dataset? That's a sensor data which u have worked also in feature selection

  • @varunraj5543
    @varunraj5543 Рік тому

    I'm following the steps as explained and training on my dataset but in all the epochs the training and validation accuracy is coming constant . My dataset is of images and has 3 classes with total of 2000 training and 200 validation images. Can someone please help in identifying where the problem is.

  • @codewithkuk
    @codewithkuk 11 місяців тому

    bhai maja agaye ,kya smajhaya h apne

  • @satyamsharma5167
    @satyamsharma5167 Рік тому

    [Very Urgent] Please tell what can we call these implementations, if applied to a disease(say covid) dataset problem: Covid detection, covid classification, covid diagnosis, covid prediction ya automated diagnosis/ automated detection. What exactly do we call it when we use/implement this code. i have no idea. Please kindly tell. People from comments also can advice. Please it is very urgent 🙏🙏

  • @RealBharatKatkade
    @RealBharatKatkade 2 роки тому +1

    Hello Sir, Please also make videos on BERT model.

  • @hafizmuhammadanwarshahzada9972
    @hafizmuhammadanwarshahzada9972 11 місяців тому

    your video are very helpful for machine and deep learning. kindly guide me can we use this model for medical images like lungs cancer etc. and plz guide for multi class classification. if you have any video on multiclass classification with the help of transfer learning, kindly share its link

  • @tanmaymore2670
    @tanmaymore2670 Рік тому

    Sir we can use unsupervised learning for label the Dataset and then perform cnn? Help me please sir

  • @athleteworld1703
    @athleteworld1703 5 місяців тому

    sir ham fine tuning ke sath data augmentation use karsakte ha

  • @manojmholagi7035
    @manojmholagi7035 2 роки тому

    I graduated in the year of 2020. Completed data science course and learning more concepts through this channel, I am still fresher, does my graduated year affect for getting job in this field?

  • @StoryScapebyShan
    @StoryScapebyShan 6 місяців тому

    completed ,. Awesome content

  • @chbilalahmad6236
    @chbilalahmad6236 Рік тому +1

    i reached 98.98% accuracy as well as val_accuracy 98.16% with Inception V3

  • @Sjcube949
    @Sjcube949 6 місяців тому

    can we apply transfer learning on tabular data?

  • @PoojaYadav-tq3yp
    @PoojaYadav-tq3yp Рік тому

    Sir I worked on kidney stones dataset, and same I faced overfitting problem . Train accuracy in 80s and test accuracy in 40s .
    But I was unable to overcome.

  • @umersiddiqui9580
    @umersiddiqui9580 10 місяців тому

    I LOVE THIS GUY

  • @anwesaroy6998
    @anwesaroy6998 Рік тому

    Hey, can you please share the link to the resources that you use while recording? For instance the scribble pad and the software.

  • @navpreetkaur9091
    @navpreetkaur9091 Рік тому

    how to use alexnet as pretrained model

  • @harikatammina6437
    @harikatammina6437 Рік тому

    Hello, I found this content useful if I have to use xception incase of VGG-16 then how to use, which layers to freeze.

  • @abhishekpawar1207
    @abhishekpawar1207 Рік тому

    I'm Doing this same thing with Pytorch plz any help me how do i fine tune

  • @UsmanKhan-hu8hw
    @UsmanKhan-hu8hw Рік тому

    can we apply the same code for the transfer learning of text sentiments ?

  • @NnColal
    @NnColal Рік тому

    Sir, Can I do transfer learning on my own custom "model" that is created by me? That is I have created and trained my own CNN model saved as xyz.keras. I now want to do transfer learning of this model on new type of data. Is that possible?

  • @xyz-pc3tl
    @xyz-pc3tl 2 роки тому

    Thanks a lot for your all efforts and hardwork you do for us🙏🙏
    Thanks for python playlist ..
    Sir , can you tell me that what i should do after completing your python playlist ( in details) ??
    Please....complete your 100 days of python playlist or you will tell us that what topics are covered in 100 days of python playlist so we can prepare..
    How can I be machine Learning Engineer or data Scientist in FACEBOOK/ GOOGLE/ AMAZON/ APPLE/ MICROSOFT ??
    Sir....please....please...reply me
    🙏🙏🙏🙏🤟🥳

  • @yourcodingerror3816
    @yourcodingerror3816 2 роки тому

    Hello sir, If has been pretrained on a text classification model, can we use this model in text generation, question answering or paragraph summary?

  • @RIFATHBINHOSSAIN-e8p
    @RIFATHBINHOSSAIN-e8p Рік тому

    Good explanation

  • @algocoholic2494
    @algocoholic2494 2 роки тому

    Unmatchable 😊

  • @muhammadaafaq691
    @muhammadaafaq691 8 місяців тому

    Our teacher told us that transfer learning is used in CNNs because initially features are somewhat similar and change in the later layers. So, my question is whether transfer learning can only be used in CNN tasks?
    Please Reply

  • @anuragshrivastava7855
    @anuragshrivastava7855 3 місяці тому +1

    NLP playlist pending

  • @aadihere-pr2wh
    @aadihere-pr2wh Рік тому

    What a great explanation. I have a question. How did you train 25000 images in Google Collab? Because When I use free Google Collab on 10000 images, all GPU units are utilized. What is the solution?

  • @hrmnde3p
    @hrmnde3p 2 роки тому +1

    Bhai Seq2Seq ki video kab aye ge ?

  • @tauseefurrahman9245
    @tauseefurrahman9245 Рік тому

    All I need .....just found in one video

  • @mishal_ashraf
    @mishal_ashraf Рік тому

    I ran the last notebook but results are not good....accuracy is 50%. I didn't make even a single change. Please have a look at the last one in which you did finetunning again. Thanks

  • @kindaeasy9797
    @kindaeasy9797 5 місяців тому

    I guess it is giving good results because it has been trained on cat dog images

  • @sandeepsasikumar701
    @sandeepsasikumar701 Рік тому

    Great video❤

  • @aniketkhot5618
    @aniketkhot5618 2 роки тому

    Sir plz continue machine learning playlist

  • @mrityunjayupadhyay7332
    @mrityunjayupadhyay7332 2 роки тому

    Great explanation