Today's Progress : I have setup all the things I needed to complete this challenge and also completed chosing the projects I will work on .
Thoughts : Hope this will be exiciting ,will help me in learning Machine Learning in a more effective way .
Today's Progress : I have completed data preparation for kaggle ml problem .
Thoughts : It is really challenging and many error occured , but I have debuged them . It was really fun .
Link of Work: Commit
Today's Progress : I have completed initial training and testing of the model and submitted to the Kaggle .
Thoughts : The score have to be improved I will try my best on it and it is really fun .
Link of Work: Commit
Today's Progress : I have completed Google ML crash course upto reduce data loss .
Thoughts : The course is really good and it contain lots of good contents . It is good for a beginner .
Link of Work: Google ML Crash Course
Today's Progress : I have done more of Google ML crash course .
Thoughts : The programing exercises are good.
Link of Work: Google ML Crash Course
Today's Progress : I have implemented Linear regression model algorithm from scratch and used it to prediction .
Thoughts : The builtin tools or models we use are also just algorithms , they are not any rocket science , learning ML become more fun when these algorithms are implemented from scratch , it was really a great experience .
Link of Work: Commit
Today's Progress : Learned about Gradient descent algorithm and its equations .
Thoughts : the concepts are really good .
Link of Work: Siraj's Gradient descent
Today's Progress : Learned and Implemented Gradient descent algorithm, Stochastic grad descent, mini batch , polynomial .
Thoughts : the concepts are really good .
Link of Work: Commit
Today's Progress : Learned and Implemented Learning curves .
Thoughts : learning curves are really helpful in determining a model is overfiting or underfiting .
Link of Work: Learning Curve
Today's Progress : Learned and Implemented Ridge regression , L2 norm , L1 Norm .
Thoughts : Regularization is a great tool for making a Model dumb(i.e putting some constraints).
Link of Work: Ridge regression
Today's Progress : Learned and Implemented Logistic regression .
Thoughts : Lots and Lots of math equation .
Link of Work: Logistic regression
Today's Progress : Learned Linear Algebra and it is continuing.
Thoughts : Having a good concept building in Maths is very important in ML .
Link of Work: Linear Algebra
Today's Progress : Continued Learning Linear Algebra .
Thoughts : Having a good concept building in Maths is very important in ML .
Link of Work: Linear Algebra
Today's Progress : Learned linear algebra and started with SVM.
Thoughts : Support vector machine is pretty amazing .
Link of Work: SVM
Today's Progress : continued with Linear algebra and SVM .
Thoughts : Gaining more and more insights , understanding more and more how deep these concepts are .
Link of Work: SVM
Today's Progress : continued with SVM , started with SVM kernel.
Thoughts : Kernel trick is really a tricky concept . Have to read more of it . And again math is very much important for ML.
Link of Work: SVM-kernel
Today's Progress : continued with SVM regression .
Thoughts : the concept of svm is really tricky .
Link of Work: SVM-Regression
Today's Progress : Revised SVM concepts
Thoughts : the concept of svm is really complex
Link of Work: SVM-Note
Today's Progress : Get started with one of the most important concept of ML Decision trees.
Thoughts : Decision trees are very efficient algorithm , hoping to learn a lot .
Link of Work: Decision Tree 1
Today's Progress : Learned more about concepts of ML Decision trees.
Thoughts : Decision tree can be visualized easily that is why they are great , also they are sensitive to the data , chance of overfiting if hyperparameters are not used properly .
Link of Work: Decision Tree 2
Today's Progress : Played with moons dataset of sklearn just practicing different concepts.
Thoughts : Coding is the real love .
Link of Work: Decision Tree 3
Today's Progress : Continued with that dataset , tweak some hyperparameter and get the 87% accuracy on a decision tree.
Thoughts : Getting used to the process of ML , less coding more thinking loving it.
Link of Work: Decision Tree 4
Today's Progress : started with ensemble learning.
Thoughts : concepts are really good.
Link of Work: Ensemble Learning
Today's Progress : Learned about concept of bagging and pasting.
Thoughts : good concepts.
Link of Work: Ensemble Learning-2
Today's Progress : Learned about out of bag evaluation and using it for validation of model.
Thoughts : ensemble learning is a beauty.
Link of Work: Ensemble Learning-3
Today's Progress : Learned about random forest classifier.
Thoughts : very good concepts.
Link of Work: Ensemble Learning-4
Today's Progress : Learned about adaptive boosting.
Thoughts : ensemble learning is full of concepts.
Link of Work: Ensemble Learning-5
Today's Progress : Learned about gradient boosting.
Thoughts : ensemble learning is full of concepts.
Link of Work: Ensemble Learning-6
Today's Progress : Learned about stacking in ensemble learning.
Thoughts : ensemble learning is full of concepts.
Today's Progress : revised ensemble learning and started playing with it with MNIST datset.
Thoughts : ensemble learning is full of concepts.
Link of Work: Ensemble Learning-7
Today's Progress : try to train it on RandomTreeClassifier.
Thoughts : machine learning is full of surprises.
Link of Work: Ensemble Learning-8
Today's Progress : Trained SVM , RandomTreeClassifier , ExtraClassifier and also trained the voting cls using the same models , randomTreeClassifier outperform all.
Thoughts : Yes RandomTreeClassifier Outperform all but still it needs enough data to perform well. and voting classifier is great overall. As you are training and compairing all the models and their results to find a good classifier.
Link of Work: Ensemble Learning -9
Ensemble Learning-10
Today's Progress : Started learning Dimensionality reduction in ML.
Thoughts : reducing runtime is a very important thing in Ml , i.e why learning about Dimensionality reduction is very important.
Today's Progress : Learned about different aspects of Principle Component analysis in dimension reduction and also how to apply them using sklearn.
Thoughts : very good concepts.
Link of Work: Dimensionality Reduction-1
Today's Progress : Learned about Kernel PCA .
Thoughts : It's complex.
Link of Work: Dimensionality Reduction-2
Today's Progress : Revised on Dimensionality reduction .
Thoughts : It's helps a lot when you revised the ideas , concepts.
Today's Progress : Compared on the MNIST dataset by reducing the dataset dimensionality and not reducing it , training RandomForest clf on this 2 datasets .
Thoughts : Its turn out that after reducing the dimension of dataset the training time increase twice of trainingtime without the reductio and also accuracy decrease by 2% . It's a surprise but we know the famous "The Curse Of Dimensonality" always unexpected
Link of Work: Dimensionality Reduction-3
Today's Progress : 1) get started tensorflow , implemented basic operations of tensorflow . 2) Implemented more operations and logistic regression using tensorflow.
Thoughts : Very much excited on implementing different operations using the most useful and famous tool for ML/AI/DL Tensorflow .
Link of Work: Tensorflow -1 Tensorflow -2
Today's Progress : implemented Gradient Descent using tensorflow.
Thoughts : it was fun .
Link of Work: Tensorflow -3
Today's Progress : Read out a new paper published by DeepMind .
Thoughts : It was really progressive thinking on the research of generalization of AI.
Link of Work: Paper -1
Today's Progress : 1) Learned more on how to make gradinent descent algo more efficient using tensorflow. 2) Implemented mini-batch gradient descent.
Thoughts : Very much excited on implementing different operations using the most useful and famous tool for ML/AI/DL Tensorflow .
Link of Work: Tensorflow -4
Today's Progress : 1) Learned about how can save your trained model and restore it whenever u need it. This explains how u can use pretrained model using tensorflow , a great feature 2) Learned how to use tensorboard . Its a really great tool .
Thoughts : Very much excited on implementing different operations using the most useful and famous tool for ML/AI/DL Tensorflow .
Link of Work: Tensorflow -5
Today's Progress : Learned about NameScope , modularity and shared variables .
Thoughts : Concepts are good.
Today's Progress : Read the paper released by google on Tensorflow , it was really awesome and full of deep knowledge and information about tensorlflow , learned a lot .
Thoughts : It helped me to clear lots of doubts , and to gain deep insight about tensorflow, everyone who wants to learn tensorflow should take a look at this paper, better than any courses.
Link of Work: Paper-2
Today's Progress : revising tensorflow , the concepts are a little tricky.
Thoughts : Concepts are good.
Today's Progress : Also revising some codes and making some previous code better of tensorflow , understanding more of tensorflow .
Thoughts : Tensorflow is kind of complex concept to understand properly , but I will spend some more time on it also because it very much important .
Link of Work: Tensorflow-6
Today's Progress : Today I learn more about tensor board and also implemented one .
Thoughts : It is a handy tool for visualization of the Model .
Link of Work: Tensorflow-7
Today's Progress : Today I learn more about how to use namescope in tensorboard .
Thoughts : It is a good concept .
Link of Work: Tensorflow-8
Today's Progress : Today I learn more about how to use modularity and shared variable in tensorflow and also implemented them .
Thoughts : It is a good concept .
Link of Work: Tensorflow-9
Today's Progress : Dome some exercises on tensorflow from online .
Thoughts : Doin exercises really help a lot .
Link of Work: Tensorflow-10
Today's Progress : Started with Artificial Neural network(ANN) , the most awesome idea since 1943.
Thoughts : ANN have been introduced in 1943 , but still we haven't come to half of its successful development. It is the proof of progress in Science can't be on a particular field , progress of science will always means the progress of the all scientific researches on all fields .
Link of Work: ANN-1
Today's Progress : Learned about perceptrons , Logical computing with neurons .
Thoughts : very good concepts.
Link of Work: ANN-2
Today's Progress : Learned about backpropagation .
Thoughts : very good concepts.
Today's Progress : I was learning and revising the Calculas .
Thoughts : learned a lot.
Link of Work: Calculas
Today's Progress : Not done much but , was doing the move 37 course these days .
Thoughts : it is a good course no doubt.
Today's Progress : Completed Move37 weeek 1 .
Thoughts : it is a good course no doubt.
Today's Progress : Revised the concepts of perceptrons , Sigmoid , ReLU , LTU , MLP.
Thoughts : Revising make the concepts clear.
Today's Progress : Read the famous paper Anybody can dance .
Thoughts : Awesome paper , great work in progress of AI .
Link of Work: Any Body Can Dance
Today's Progress : I was revising topics like Gradient Descent , Linear Regression , Logistic regression etc.
Thoughts : revising make concepts more clearer.
Today's Progress : Doing Move 37 2nd week.
Thoughts : very good concepts.
Today's Progress : Done Move 37's advanced bellman and Dynamic equation .
Thoughts : very good concepts .
Today's Progress : Implemented Multilayer Neural network using tensorflow deep neural network.
Thoughts : The performance is really great.
Link of Work: Commit
Today's Progress : I have implemented a custom deep neural network algorithm from scratch and trained it on the mnist dataset .
Thoughts : It was a lot of learning.
Link of Work: Commit-1 Commit-2 Commit-3
Today's Progress : 1) Brain stormed some questions on Deep neural network. 2) trying to get more accuracy than 98% , on the mnist dataset using DNN .
Thoughts : Still have to do more tunning and tweak of parameters.
Link of Work: Commit