Kemal Erdem Blog RSS Feed

Step by Step visual introduction to Diffusion Models.

Wed, 01 Nov 2023 00:00:00 GMT

What is a diffusion model? The idea of the diffusion model is not that old. In the 2015 paper called “Deep Unsupervised Learning using Nonequilibrium Thermodynamics” [1], the Authors described it like this: The essential idea, inspired by non…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

XAI Methods - Guided GradCAM

Mon, 18 Apr 2022 00:00:00 GMT

What is Guided GradCAM method? Guided GradCAM is a method created by Selvaraju et al. [1] which combines both GBP and GradCAM (also created by the same authors). To explain the idea of Guided GradCAM, we have to split it into separate components…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

XAI Methods - Noise Tunnel

Mon, 18 Apr 2022 00:00:00 GMT

What is the Noise Tunnel? Noise Tunnel [1] is not an attribution method but a technique that improves the accuracy of attribution methods. It combines SmoothGrad [2], SmoothGrad-Square (unpublished version of the SmoothGrad), and VarGrad [3] and…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

XAI Methods - Integrated Gradients

Mon, 18 Apr 2022 00:00:00 GMT

What is Integrated Gradients method? Integrated Gradients (IG) [1] is a method proposed by Sundararajan et al. that is based on two axioms: Sensitivity and Implementation Invariance. Authors argue that those two axioms should be satisfied by all…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

XAI Methods - Guided Backpropagation

Tue, 22 Feb 2022 00:00:00 GMT

What is Guided Backpropagation? Guided Backpropagation (GBP) [3] is an approach designed by Springenberg et al., relying on the ideas of Deconvolution [1] and Saliency [2]. Authors argue that the approach taken by Simonyan et al. [2] has an issue…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

XAI Methods - Deconvolution

Mon, 21 Feb 2022 00:00:00 GMT

What is Deconvolution? The idea of Deconvolution [2] comes from the work of Zeiler et al. [1] about Deconvolutional Networks (deconvnets}. Deconvnets are designed to work similar to convolutional networks but reverse (reversing pooling component…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

XAI Methods - Saliency

Mon, 21 Feb 2022 00:00:00 GMT

What is Saliency and why the same is confusing? Saliency [1] is one of the first attribution methods designed to visualize the input attribution of the Convolutional Network. Because the word saliency is often related to the whole approach to display…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

XAI Methods - The Introduction

Tue, 12 Oct 2021 00:00:00 GMT

Explainable Artificial Intelligence Explainable Artificial Intelligence (XAI) is one of the youngest and one of the fastest developing branches of the field. The objective of the XAI method is to provide an explanation for the deep learning model…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

Introduction to Attention Mechanism

Wed, 12 May 2021 00:00:00 GMT

The attention mechanism is one of the most important inventions in Machine Learning, at this moment (2021) it’s used to achieve impressive results in almost every field of ML, and today I want to explain where it came from and how it works. Before…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

Understanding Positional Encoding in Transformers

Mon, 10 May 2021 00:00:00 GMT

What is positional encoding? As I’ve explained in “Introduction to Attention Mechanism”, attention doesn’t care about the position of the inputs. To fix that problem we have to introduce something called Positional Encoding. This encoding is covered…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

Measuring XAI methods with Infidelity and Sensitivity

Tue, 16 Mar 2021 00:00:00 GMT

Explainable Artificial Intelligence (XAI) is getting more and more popular every year. Also every year we have more methods that aim to explain how our models work. The only problem with most of those approaches is comparison. We have two types of…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

Animal recognition with Siamese Networks and Mean Embeddings

Thu, 25 Feb 2021 00:00:00 GMT

When hearing about Siamese Networks you probably think about “Face Recognition”. That’s the most common use of those types of networks. We were trying to do sth else, recognize animals based only on top-view camera footage. If you’re not interested…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

DengAI - Data preprocessing

Mon, 03 Aug 2020 00:00:00 GMT

This article is based on my entry into DengAI competition on the DrivenData platform. I’ve managed to score within 0.2% (14/9069 as on 02 Jun 2020). In this article, I assume that you’re already familiar with DengAI - EDA. You don’t have to read it…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

DengAI - How to approach Data Science competitions? (EDA)

Fri, 31 Jul 2020 00:00:00 GMT

This article is based on my entry into DengAI competition on the DrivenData platform. I’ve managed to score within 0.2% (14/9069 as on 02 Jun 2020). Some of the ideas presented here are strictly designed for competitions like that and might not be…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

Evolution of Extreme Learning Machines

Thu, 23 Jul 2020 00:00:00 GMT

Note! This is just a major overview of the ELM evolution. It doesn’t include all possible versions and tweaks done to ELMs through the years. What is ELM? ELM (Extreme Learning Machines) are feedforward neural networks. “Invented” in 2006 by G. Huang…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

Finding correlations in time series data

Sun, 28 Jun 2020 00:00:00 GMT

Problem description If you’ve ever worked with data analysis it’s highly likely that you know about the idea of data correlation. There are many ways of calculating correlation within your data, and most of them are already implemented in popular…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

Introduction to Extreme Learning Machines

Wed, 13 May 2020 00:00:00 GMT

What is ELM? ELM (Extreme Learning Machines) are feedforward neural networks. “Invented” in 2006 by G. Huang. As said in the original paper: this algorithm tends to provide good generalization performance at extremely fast learning speed Hence the…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

t-SNE clearly explained

Mon, 13 Apr 2020 00:00:00 GMT

What is t-SNE? Many of you already heard about dimensionality reduction algorithms like PCA. One of those algorithms is called t-SNE (t-distributed Stochastic Neighbor Embedding). It was developed by Laurens van der Maaten and Geoffrey Hinton in 200…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

Making TensorflowJS work faster with WebWorkers.

Mon, 17 Feb 2020 00:00:00 GMT

The issue If you’ve ever worked with TensorflowJS you probably know that running data through models (especially image processing) takes a long time. Because JavaScript is single-threaded, that poses a huge problem for a site responsiveness. Every…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

Understanding Region of Interest - Part 2 (RoI Align)

Sun, 09 Feb 2020 00:00:00 GMT

This is the second part of RoI Series. It doesn’t include an introduction about what RoI is, so if you didn’t read the First Part it might be hard to follow this one. Why do we want to modify RoI Pooling? As you remember from the first part of the…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

Understanding Region of Interest (RoI Pooling)

Thu, 06 Feb 2020 00:00:00 GMT

We’re going to discuss original RoI pooling described in Fast R-CNN paper (light blue rectangle on the image above). There is a second and a third version of that process called RoIAlign and RoIWarp. If you’re interested in those two please check out…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

How to create Tensorflow 2 Sequence Dataset from scratch

Sat, 28 Dec 2019 00:00:00 GMT

The problem Modern Machine Learning solutions require a huge amount of data, that’s definitely the case when working with image recognition/object detection. Because of that, we need to create more and more complex datasets to teach our models. At…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

JSF*ck - Does code weirdness affect execution performance?

Thu, 07 Nov 2019 00:00:00 GMT

This post is created base on Martin Kleppe’s (@aemkei) work about converting JS to only six basic characters and still keeping it executable. It is a response to comment from Jonathan Means. For those who are reading this out of the context, I’m…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

React - state management without libraries (with hooks) PART 2

Sun, 15 Sep 2019 00:00:00 GMT

It might be beneficial to have CodeSandbox editor opened when reading this blog post. This is a second part of “State management without libraries”. If you didn’t read first one, it might be difficult to follow this one, so please check out Part…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

React - state management without libraries (with hooks)

Fri, 06 Sep 2019 00:00:00 GMT

It might be beneficial to have CodeSandbox editor opened when reading this blog post. Why don’t you want to use state manager anyway? State managers (like Redux or MobX) are really great but imho they are just to complex for small apps. What is a…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

A modern guide to Events in JavaScript

Mon, 19 Aug 2019 00:00:00 GMT

What are Events? Events are objects and implement the Event interface Those objects are dispatched by the user agent (or the application itself) and handled by the Event Listener. Events are responsible for handling user interactions or network…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

React performance tricks

Sat, 10 Aug 2019 00:00:00 GMT

This post requires a basic knowledge about Shapes and Inline Cache. If you didn’t read V8 function optimization, it might be difficult to follow this one. Every time someones talk about “performance” in React it’s usually about virtual DOM. It is an…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

Can JSON.parse() be performance improvement?

Wed, 07 Aug 2019 00:00:00 GMT

This post requires a basic knowledge about Shapes and Inline Cache. If you didn’t read V8 function optimization, it might be difficult to follow this one. Problem Creating a copy of an object is common practice in JS. You’ve probably done that while…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

V8 function optimization

Tue, 06 Aug 2019 00:00:00 GMT

Setup (optional) If you don’t want to run it on your computer, please skip to this section First, we need to install V8 to be able to run it without whole package (node or web browser). I’ve created gist describing the process for Linux users. https…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

Is Symbol really useful?

Wed, 24 Jul 2019 00:00:00 GMT

Quick recap, what are those “Symbols” anyway? primitive value that represents a unique, non-String Object property key That’s Symbol definition from current spec. But what does it mean? You probably know other primitive types in JS (Undefined, Null…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)

How JS becomes multi threaded?

Thu, 11 Jul 2019 00:00:00 GMT

This is advance staff, you don’t really need it to be great developer, so don’t be bothered if you don’t get everything at the beginning. Reading this makes you more aware than maybe 90% of all JS devs :) Throughout the years, JavaScript was…

(This is an article posted to my blog at erdem.pl. You can read it online by clicking here.)