Portfolio

About Me

I've been learning about AI and its applications for around one year. I've got an internship at an AI startup based on Tokyo where I've been sharping my skills in machine learning. I've been implementing several states-of-the-art AI models and publishing them in my Github account.

What I like the most in Machine Learning is what we call Computer Vision. I love to make much use of AEs, VAEs and GANs for dealing with pictures like Denoising, Super-resolution, Anomaly Detection, Colorization and Segmentation.

I am keeping myself refreshed with up-to-date AI models in Computer Vision. I am looking forward to understand and master Detection Models and GANs. I would like to make use of Detection Models like Yolo, SSD and Mask-RCNN for Autonomous Driving and Tumors Detection in MedTech.

Latest Projects

New

Bao-Classifier

For now, this project includes the implementation of the following models in Tensorflow 2.x:

Family	Models	Reference
AlexNet	AlexNet	paper
ZFNet	ZFNet	paper
LeNet	LeNet4, LeNet5	paper
VGG	VGG11, VGG13, VGG16, VGG19	paper
SqueezeNet	SqueezeNetV1, SqueezeNetV2, SqueezeNetV3	paper
ResNet	Resnet18, Resnet34, Resnet50, Resnet101, Resnet152	paper

Back my project

Segmentation

I implemented the honorable U-NET and successfully ran it on Cityscapes Dataset. I was surprised that the model worked out well although it was trained in few epochs. I hope I can find the time to implement other models and apply them on MedTech tasks like detecting tumors in CT images.

Denoising

This project is a simple denoising AE & VAE written in tensorflow 2. For the moment, I am experiencing the efficiency of AEs and VAEs models to remove the noise from MNIST and Flowers datasets. To find out more, please refer to the following repositories in my Github account.

Denoising AE
Vanilla VAE
Stacked VAE
Conv VAE

Colorization & Super-resolution

Japanese Anime is just wonderful, I started experiencing to see how far the robustness of Machine Learning can go on coloring anime sketches. I am experiencing several models for it: AutoEncoders, U-NET, Plain Convolutional Network... These models may work on other tasks as well:

GrayScale	----> Color
Season	<---> Season
Day	<---> Night
Images Super-resolution

For the moment, I tried a very basic model to see how the things will work out, the result of this model is shown in the following picture. You can check my updated Github account for seeing detailed information. anime colorization

ML Tutorials

Machine Learning is expanding rapidly nowadays, I have been making some tutorials as personal notes and for whoever wanna know more about AI. The tutorials include simple-to-hard examples on the following libraries:

OpenCV	is a magnificient library for anyone who is interested in Computer Vision.
NumPy	is a library for supporting large, multi-dimensional arrays and matrices.
Pandas	is a Python Data Analysis library. I can say it is the SQL for Data Analysts
Matplotlib	Matplotlib is a plotting library for the Python and its numerical mathematics extension NumPy.
Tensorflow 2.0	My favorite ML library. I also wrote some simple tutorials for it, please find out Simple Classifiers & Simple Regressors

Other AI Projects

Note: All the following algorithms were implemented from scratch in Python 3

Metaheuristics

I've implemented some popular Metaheuristics algorithms for finding the minimum of a given mathematical function.
The following algorithms were all implemented:

For more info, please check them out in my Github account

Metaheuristics 2D Metaheuristics 3D

Evolutionary Algorithms

This category of algorithms is actually a subset of Metaheuristics. I've successfully implemented 3 Evolutionary Algorithms that can find the minimum of a given 2D or 3D mathematical function:

For more info, please check them out in my Github account

Genetics 2D Genetics 3D

Numerical Analysis

Nummerical Analysis is a vast and huge branch in Mathematics. During my journey in learning about ML, I came across some Numerical Analysis based algorithms (e.g. Gradient Descent) which were in fact used in developing the core of some well-known ML Libraries like Tensorflow and PyTorch. I could implement some NA algorithms in Python for finding the root of a given 2D funtion:

For more info, please check them out in my Github account

Numerical Analysis

SVM, Perceptron and Backpropagation

In the journey of understanding ML algorithms, I could not grasp how the stuff really works until I started trying to implement some core algorithms. The hard one is Backpropagation and Perceptron. SVM seems hard, but thanks to Pegasos Algorithms, it turns out that it is easy to implement.
I implemented some algorithms for solving only the linear classification problem:

In my Github account, you can find more information about:
SVM, SDCA and SGD here

Perceptron and Backpropagation here

My GitHub

You can check my GitHub contribution graph calendar using my GitHub Calendar widget.

You can also review my repositories using my GitHub Repositories widget.

My Articles & Tutorials:

Lesson 1 : NumPy Basics (En)

This tutorial include some very simple exercises about numpy used in array ,some of them you can have many solutions to solve ,it helps you to get familiar with numpy.

Lesson 2 : NumPy Matrices (En)

After getting familiar with numpy array ,we start studying about numpy used in matrices.This part will show you some basic operations about numpy matrices, for example : get the sum, min, max of a matrix or many matrices.

Lesson 3 : Applying Numpy to AI (En)

This part is very important for machine learning ,it introduces convolution, pooling , padding and other functions for dealing with images, it's very important for machine learning.

Lesson 4 : Pandas Tutorials (En)

This part includes quick start to pandas , the operations about table , rows, columns, items, and how to deal with files by using pandas, at the end it has some exercises for practicing .

Lesson 5 : OpenCv Tutorials (En)

This part you can learn many skills about dealing with pictures, for example: read images, change the size, color and add filters. OpenCv provides a library which you can find many tools to edict images easily.

Lesson 6 : Matplotlib Tutorials (En)

By studying this part you can learn how to use matplotlib to draw many different Simple Plots, Subplots, Graphs, Pie Chart and 3D Plot.

Lesson 7 : Object Detection (En)

Object detection is a computer technology related to computer vision and image processing that deals with detecting instances of semantic objects of a certain class (such as humans, buildings, or cars) in digital images and videos.You can start knowing about how it works and it's principle .

Lesson 8 : Data preparation for AutoEncoder (En)

This is a simple tutorial about preparing the training data and labels for colorization.

Lesson 9 : Python Tutorials (En)

Here is a Glossary of the python lessons I have Prepared

Lesson 10 : About VGG networks (Cn)

In this passage I introduce the vgg networks from different aspects ,such as the structures ,the kinds of vgg networks and their difference.

Bao Jiarong

I am a Machine Learning Programmer & Data Analyst.
This page is my professional portfolio in which I summarize and publish my work in the field of ML.