Deep learning sneak peek
I try to explain various important terms of deep learning and machine learning. I will write this sort of tutorial for helping myself to build a clear understanding. If anyone get helped reading this It would be grateful for me
Deep learning (Basics)
| Part | Topic | Git | Colab |
|---|---|---|---|
| 01 | āĻā§āĻ¨ā§āĻ¸āĻ°āĻĢā§āĻ˛ā§āĻ° āĻĒā§āĻ°āĻĨāĻŽāĻŋāĻ āĻ§āĻžāĻ°āĻ¨āĻž, āĻā§āĻ¨ā§āĻ¸āĻ°, āĻāĻŋāĻā§ āĻŽā§āĻ¯āĻžāĻĨ āĻ āĻĒāĻžāĻ°ā§āĻļāĻ¨ | link | x |
| 02 | āĻā§āĻ°āĻžāĻĄāĻŋā§ā§āĻ¨ā§āĻ-āĻā§āĻĒ, āĻšāĻžāĻā§āĻžāĻ° āĻ āĻĄāĻžāĻ° āĻĄā§āĻ°āĻŋāĻā§āĻāĻŋāĻ , āĻā§āĻ°āĻžāĻĄāĻŋā§ā§āĻ¨ā§āĻ-āĻā§āĻĒā§āĻ° āĻāĻŋāĻā§ āĻŦāĻŋāĻļā§āĻˇ āĻŦā§āĻ¯āĻŦāĻšāĻžāĻ° | link | x |
| 03 | āĻ˛āĻŋāĻ¨āĻŋā§āĻžāĻ° āĻ°āĻŋāĻā§āĻ°ā§āĻ¸āĻ¨, āĻĒā§āĻ˛āĻ āĻ˛āĻžāĻ°ā§āĻ¨āĻŋāĻ āĻāĻžāĻ°ā§āĻ | link | x |
| 04 | āĻ¸āĻĢāĻāĻŽā§āĻ¯āĻžāĻā§āĻ¸ āĻĢāĻžāĻāĻļāĻ¨ āĻāĻŋ āĻāĻŦāĻ āĻā§āĻ¨ āĻāĻžāĻ āĻāĻ°ā§ | link | x |
| 05 | āĻŽā§āĻļāĻŋāĻ¨ āĻ˛āĻžāĻ°ā§āĻ¨āĻŋāĻā§ā§ āĻāĻžāĻāĻĒ - ā§§ āĻ āĻāĻžāĻāĻĒ - ā§¨ āĻāĻ°āĻ° āĻāĻŋ ? | link | x |
Deep learning (Computer vision)
| Part | Topic | Git | Colab |
|---|---|---|---|
| 01 | āĻāĻŽā§āĻ āĻĄāĻžāĻāĻž āĻ˛ā§āĻĄāĻŋāĻ, āĻĢā§āĻ˛āĻžāĻā§āĻžāĻ° āĻĄāĻžāĻāĻžāĻ¸ā§āĻ, āĻĒā§āĻ˛āĻ āĻāĻŽā§āĻ | link | colab |
| 02 | āĻāĻŽā§āĻ āĻĄāĻžāĻāĻž āĻĒāĻžāĻāĻĒāĻ˛āĻžāĻāĻ¨, Keras ImageDataGenerator, āĻā§āĻžāĻ¨-āĻšāĻ āĻāĻ¨āĻā§āĻĄāĻŋāĻ | link | colab |
| 03 | āĻāĻŽā§āĻ āĻĄāĻžāĻāĻž āĻĒāĻžāĻāĻĒāĻ˛āĻžāĻāĻ¨, tf.data , map āĻĢāĻžāĻāĻļāĻ¨ā§āĻ° āĻŦā§āĻ¯āĻžāĻŦāĻšāĻžāĻ° | link | colab |
| 04 | tf.data āĻ ImageDataGenerator āĻāĻ° āĻŽāĻ§ā§āĻ¯ā§ āĻ¤ā§āĻ˛āĻ¨āĻž, āĻā§āĻ°ā§āĻ¨āĻŋāĻ āĻāĻžāĻāĻŽ āĻ¸ā§āĻĒā§āĻĄ āĻāĻĒ āĻāĻ°āĻž, āĻĄāĻžāĻāĻž āĻā§āĻ¯āĻžāĻļāĻŋāĻ | link | colab |
| 05 | āĻāĻŽā§āĻ āĻĄāĻžāĻāĻž āĻĒāĻžāĻāĻĒāĻ˛āĻžāĻāĻ¨ tf.keras.utils.Sequence | link | colab |
| 07 | āĻĢā§āĻ˛āĻžāĻā§āĻžāĻ° āĻāĻŽā§āĻ āĻā§āĻ˛āĻžāĻ¸āĻŋāĻĢāĻŋāĻā§āĻļāĻ¨, āĻ¸āĻžāĻŦāĻā§āĻ˛āĻžāĻ¸ āĻŽāĻĄā§āĻ˛, tf.keras.utils.Sequence | link | colab |
| 08 | āĻāĻžāĻ¸ā§āĻāĻŽ āĻ˛ā§ā§āĻžāĻ° āĻāĻŋ ? āĻāĻŋāĻāĻžāĻŦā§ āĻāĻžāĻ¸ā§āĻāĻŽ āĻ˛ā§ā§āĻžāĻ° āĻĄāĻŋāĻāĻžāĻāĻ¨ āĻāĻ°āĻ¤ā§ āĻšā§ | link | X |
Audio
| Part | Topic | Git | Colab |
|---|---|---|---|
| 01 | āĻ āĻĄāĻŋāĻ āĻĢāĻžāĻāĻ˛ āĻ°āĻŋāĻĄ, āĻĢāĻŋāĻāĻžāĻ°(FT, STFT, Spectrogram, MFCC) āĻāĻŋāĻā§ā§āĻžāĻ˛āĻžāĻāĻ | link | colab |
| 02 | āĻ¸ā§āĻĒāĻŋāĻ āĻā§ āĻā§āĻā§āĻ¸āĻ āĻ¸āĻŋāĻ¸ā§āĻā§āĻŽ āĻĄāĻŋāĻāĻžāĻāĻ¨ āĻ āĻā§āĻ¨ āĻāĻŽāĻ°āĻž āĻ¸ā§āĻĒā§āĻā§āĻā§āĻā§āĻ°āĻžāĻŽ āĻŦā§āĻ¯āĻŦāĻšāĻžāĻ° āĻāĻ°āĻŋ? | link | - |
Deep learning (NLP)
| Part | Topic | Git | Colab |
|---|---|---|---|
| 01 | āĻā§āĻā§āĻ¨āĻžāĻāĻāĻžāĻ° āĻāĻŋ āĻāĻŦāĻ āĻāĻāĻžāĻ° āĻŦāĻŋāĻāĻŋāĻ¨ā§āĻ¨ āĻĒā§āĻ°ā§ā§āĻ | link | colab |
| 02 | āĻā§āĻā§āĻ¨āĻžāĻāĻāĻžāĻ° āĻāĻ° āĻāĻ°āĻ āĻāĻŋāĻā§ āĻŦā§āĻ¯āĻŦāĻšāĻžāĻ° | lin k | colab |
ML Ops
| SL | Topic | Link |
|---|---|---|
| 01 | Rule of Machine Learning | link |
| 02 | MLOPS 101: Tips, Tricks and Best Practices- Vladimir Osin PyData Eindhoven 2021 | link |
| 03 | Hands on MLOps using CML tool | link |
| 04 | ML development process | link |
ML data visualization
CNN
- Machine Learning Foundations by google developer playlist
- Implement various CNN youtube playlist by MIT
- CNN Architectures - implementations | MLT
RNN
Transformers
The loss functions
NLP
- Building models with tf.text (TF World '19)
- Natural Language Processing (NLP) Zero to Hero - Play list
Datasets
Courses
- Intro to tensorflow for deeplearning (Udacity)
- Full stack deeplearning
- Software engineering with ML (udacity)
- Machine Learning Crash Course by Google
- TensorFlow, Keras and deep learning, without a PhD
- Deep learing by stanford - CS230
- MIT deep learning lecture
- Intro to deep learning kaggle mini course
- Kaggle free data science course
Bangla ML/DL resource:
- āĻšāĻžāĻ¤ā§āĻāĻ˛āĻŽā§ āĻĒāĻžāĻāĻĨāĻ¨ āĻĄāĻŋāĻĒ āĻ˛āĻžāĻ°ā§āĻ¨āĻŋāĻ
- āĻ¸āĻšāĻ āĻŦāĻžāĻāĻ˛āĻžā§ 'āĻŦāĻžāĻāĻ˛āĻž' āĻ¨ā§āĻ¯āĻžāĻāĻžāĻ°āĻžāĻ˛ āĻ˛ā§āĻ¯āĻžāĻā§āĻā§ā§ā§āĻ āĻĒā§āĻ°āĻ¸ā§āĻ¸āĻŋāĻ (āĻāĻ¨āĻāĻ˛āĻĒāĻŋ)
- āĻŦāĻžāĻāĻ˛āĻžā§ āĻŽā§āĻļāĻŋāĻ¨ āĻ˛āĻžāĻ°ā§āĻ¨āĻŋāĻ
- āĻĄāĻŋāĻĒ āĻ˛āĻžāĻ°ā§āĻ¨āĻŋāĻ āĻ āĻāĻ°ā§āĻāĻŋāĻĢāĻŋāĻļāĻŋā§āĻžāĻ˛ āĻ¨āĻŋāĻāĻ°āĻžāĻ˛ āĻ¨ā§āĻāĻā§āĻžāĻ°ā§āĻ
- āĻŦāĻžāĻāĻ˛āĻžā§ āĻŦā§āĻ¯āĻžāĻ¸āĻŋāĻ āĻĄāĻžāĻāĻž āĻ¸āĻžā§ā§āĻ¨ā§āĻ¸ āĻļā§āĻāĻžāĻ° āĻā§āĻ°ā§āĻ¸
Hyper params tunes
- Effect of batch size on training dynamics
- Determining optimal batch size
- Hyperparameter Importance | PyTorch Developer Day 2020
Machine learning in production
- Rules of ML by google
- 5 Steps to take your model in production
- Building the Software 2 0 Stack (Andrej Karpathy)
- A Recipe for Training Neural Networks (Andrej Karpathy)
- Engineering Practices for Software 2.0 (PyTorch Developer Conference)
- Machine Learning Interviews: Lessons from Both Sides - FSDL
- Troubleshooting Deep Neural Networks
- Machine Learning System Design (Chip Huyen)
- Machine Learning Systems Design (Chip Huyen) CS 329S
- ML Interview book (chip Huyen)
- Real Time Machine Learning Challenges and Solution (Chip Huyen)
- Data Distributio Shifts and monittorig (Chip Huyen)
Miscellaneous
- data-science-with-dl-nlp-advanced-techniques
- 37-reasons-why-your-neural-network-is-not-working
- https://cs231n.github.io/
- tf-keras-rnn-ctc-example
- Keras debugging tips
- TF2 mixed presition training speed up
- Effect of activation fucntion
- tf2 object detection api
- Eat tensorlfow in 30 days
- Tensorflow examples
- TensorFlow Fall 2020 updates: Keynote & whatâs new since TF2.2
- Explainable Ai
- The NLP Index
- Machine Learning Glossary
- Beam search, How it works
- Nvidia tao toolkit for developerhttps://developer.nvidia.com/tao-toolkit