Posts by Year



CBAM: Convolutional Block Attention Module

4 minute read

“CBAM: Convolutional Block Attention Module” proposes a simple and effective attention module for CNN which can be seen as descendant of Sqeeze and Excitatio...

Re-ID done right

5 minute read

“Re-ID done right: towards good practices for person re-identification” proposes a different approach to use deep network on person re-identification task. I...


5 minute read

“Deep image retrieval: learning global representations for image search” proposes an approach for instance-level image retrieval. It was presented in the ECC...


4 minute read

“Squeeze-and-Excitation Networks” suggests simple and powerful layer block to improve general convolutional neural network. It was presented in the conferenc...

Regularization and Optimization

1 minute read

This post is a summary and paper skimming on regularization and optimization. So, this post will be keep updating by the time.

Scale-Invariant Feature Transform (SIFT)

4 minute read

Scale-Invariant Feature Transform (SIFT) is an old algorithm presented in 2004, D.Lowe, University of British Columbia. However, it is one of the most famous...

Image Retrieval

4 minute read

This post is a summary and paper skimming on image retrieval related research. So, this post will be keep updating by the time.

Rotation Invariance & Equivariance

13 minute read

This post is a summary and paper skimming on rotation invariance and equivariance related research. So, this post will be keep updating by the time.

Detection & Segmentation

3 minute read

This post is a summary and paper skimming on detection and segmentation related research. So, this post will be keep updating by the time.

Inception-v4 and Inception-ResNet

1 minute read

“Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning” is an advanced version of famous vision model ‘inception’ from Google. It...

Network In Network

4 minute read

“Network In Network” is one of the most important study related convoutional neural network because of the concept of 1 by 1 convolution and global average p...


7 minute read

“Hide-and-Seek: Forcing a Network to be Meticulous for Weakly-supervised Object and Action Localization” proposed a weakly-supervised framework to improve ob...

Loss Functions

3 minute read

When we train a deep learning model, we need to set a loss function for minimizing the error. The loss function indicates how much each variable contributes ...

Attention Is All You Need

5 minute read

The paper “Attention is all you need” from google propose a novel neural network architecture based on a self-attention mechanism that believe to be particul...


15 minute read

It is always important what kind of optimization algorithm to use for training a deep learning model. According to the optimization algorithm we use, the mod...

Vision Technique

2 minute read

Research on several vision techniques such as pixel difference and optical flow.

Approximate Inference

3 minute read

Approximate inference methods make it possible to learn realistic models from big data by trading off computation time for accuracy, when exact learning and ...


9 minute read

Autoencoder is an artificial neural network used for unsupervised learning of efficient codings. The aim of an autoencoder is to learn a representation (enco...

Binary Search

3 minute read

Learn about Binary Search which is a simple and very useful algorithm whereby many linear algorithms can be optimized to run in logarithmic time.

Dynamic Programming

4 minute read

Learn about Dynamic Programming which is a famous and important algorithm for solving problems.

AI Related Terms

11 minute read

This post will be about artificial intelligence related terms including linear algebra, probability distribution, machine learning and deep learning

Generative Adversarial Networks (GAN)

4 minute read

Generative Adversarial Networks (GAN) is a framework for estimating generative models via an adversarial process by training two models simultaneously. A gen...


3 minute read

Regularization is important technique for preventing overfitting problem while training a learning model.


4 minute read

Learn about merge-sort, quick-sort, other sorting algorithms and their running time.

Batch Normalization

5 minute read

‘Batch Normalization’ is an basic idea of a neural network model which was recorded the state-of-the art (4.82% top-5 test error) in the ImageNet competition...


5 minute read

Learn about graph, graph representations, graph traversals and their running time.


12 minute read

Learn about probability which are the basics of artificial intelligence and deep learning.


4 minute read

Learn about tree, tree traversal, binary heap, trie and their running time.

Stack and Queue

3 minute read

Learn about stack, queue, dequeue, its implementation and running time.

Hash Table

3 minute read

Learn about hash table, hash function, hash code, its implementation and running time.

Recurrent Neural Network

5 minute read

Learn the basics about Recurrent Neural Network (RNN), its detail and case models of RNN.

YOLO 9000

less than 1 minute read

‘YOLO9000: Better, Faster, Stronger’ proposed an improved version of YOLO which was presented at IEEE Conference on Computer Vision and Pattern Recognition i...

You Only Look Once (YOLO)

4 minute read

‘You Only Look Once: Unified, Real-Time Object Detection’ (YOLO) proposed an object detection model which was presented at IEEE Conference on Computer Vision...


Faster R-CNN

3 minute read

Faster R-CNN is an object detecting network proposed in 2015, and achieved state-of-the-art accuracy on several object detection competitions.

Capsule Network

5 minute read

Cpasule Network is a new types of neural network proposed by Geoffrey Hinton and his team and presented in NIPS 2017. As Geoffrey Hinton is Godfathers of Dee...


8 minute read

TensorFlow is a machine learning framework that Google created and used to design, build, and train deep learning models. It supports complex and heavy numer...


less than 1 minute read

Keras is a high-level neural networks API, written in Python and capable of running on top of TensorFlow, CNTK, or Theano. This article is about summary and ...

Feed-Forward Neural Network (FFNN)

1 minute read

The Feed-Forward Neural Network (FFNN) is the simplest and basic artificial neural network we should know first before talking about other complicated networ...

Activation functions

8 minute read

Let’s talk about activation function in artificial neural network and some questions related of it.

Multi-Speaker Tacotron in TensorFlow

3 minute read

Today, I am going to introduce interesting project, which is ‘Multi-Speaker Tacotron in TensorFlow’. It is a speech synthesis deep learning model to generate...

Theano development ends

less than 1 minute read

Theano will not be maintained after the 1.0 release, announced by the MILA group.

Convex optimization problem

2 minute read

When we solve machine learning problem, we have to optimize a certain objective function. One of the case of it is convex optimization problem which is a pro...

References about eXplainable AI(XAI)

1 minute read

The eXplainable Artificial Intelligence (XAI) is an artificial intelligence model that is able to explain its decisions and actions to human users.