Collection of resources which I found interesting while learning ML
- “Grounding Dino” is a zero shot learning model. Something like Image transformers. It does not require labelling while training. Link : https://blog.roboflow.com/grounding-dino-zero-shot-object-detection/
- Autoencoders can be used for image noise filtering and anomaly detection from data. Link : https://blog.roboflow.com/grounding-dino-zero-shot-object-detection/
- For Anomaly Detection : https://www.youtube.com/watch?v=6S2v7G-OupA
- For tracking object and counting unique object Deep SORT is used. Link: https://github.com/abewley/sort
- BERT and GPT. : https://www.youtube.com/watch?v=9Y7f4j396hI
- GPT is mainly focused on “Decoding” part.
- BERT is focused on “Encoding” part.
- Important steps regarding tflite : https://www.tensorflow.org/lite
- Take a look into YOLO-NAS : https://blog.roboflow.com/yolo-nas-how-to-train-on-custom-dataset/
- Snapshot Ensemble Deep Learning : https://machinelearningmastery.com/snapshot-ensemble-deep-learning-neural-network/
- Selective search and other search techniques in deep learning : https://learnopencv.com/selective-search-for-object-detection-cpp-python/#:~:text=Selective%20Search%20is%20a%20region,texture%2C%20size%20and%20shape%20compatibility.
- High precision vs high recall : https://chat.openai.com/share/4d2aabd1-ab0a-4532-8eda-f727d755c22f
- Working principle and comparison of object detection algorithms : https://towardsdatascience.com/r-cnn-fast-r-cnn-faster-r-cnn-yolo-object-detection-algorithms-36d53571365e
- Graph convolutional network : https://towardsdatascience.com/understanding-graph-convolutional-networks-for-node-classification-a2bfdb7aba7b
- Data drift and concept drift in machine learning : https://towardsdatascience.com/machine-learning-in-production-why-you-should-care-about-data-and-concept-drift-d96d0bc907fb
- Internal Covariate Shift in Machine Learning : https://medium.com/analytics-vidhya/internal-covariate-shift-an-overview-of-how-to-speed-up-neural-network-training-3e2a3dcdd5cc
- For Error Analysis in ML model: https://erroranalysis.ai/, https://fairlearn.org/
- Experiment Tracking : https://neptune.ai/blog/ml-experiment-tracking
- Best practices with Tensorflow : https://cs230.stanford.edu/blog/datapipeline/#best-practices
- Feedback loop for Active Learning labelling : https://www.linkedin.com/pulse/feedback-loop-machine-learning-labeling-data-sanchit-tiwari
- Analyzing log files framework : https://www.elastic.co/logstash
- Know more about Elastic Search, seems interesting : https://www.knowi.com/blog/what-is-elastic-search/
- How Ber tokenizer works : https://www.analyticsvidhya.com/blog/2021/09/an-explanatory-guide-to-bert-tokenizer/
- "Concept Drift VS Data Drift in Machine Learning" : https://nulltoinfinity.hashnode.dev/concept-drift-and-data-drift-in-machine-learning
- It is important to understand data distribution before using the dataset. And also find out if the dataset is good/fair. Facet is an interesting library/ software. https://pair-code.github.io/facets/
- A feature column should not be included in prediction if it is not significant enough. Assume a binary classification model, If the data distribution of a feature for class 0 is similar to the data distribution for class 1 then the feature column might not bear much significance.
- Data collection, processing guideline : https://pair.withgoogle.com/chapter/data-collection/
- To inspect model after training : https://pair-code.github.io/what-if-tool/
- To understand NLP models after training : https://pair-code.github.io/lit/
- Evolutionary Algorithms : https://medium.com/mlearning-ai/why-you-should-implement-evolutionary-algorithms-in-your-machine-learning-projects-ee386edb4ecc
- Understanding SHAP for explainable AI : https://meichenlu.com/2018-11-10-SHAP-explainable-machine-learning/
- Python best practices : https://dagster.io/blog/python-project-best-practices
- LSTM return sequences and Return States : https://machinelearningmastery.com/return-sequences-and-return-states-for-lstms-in-keras/