---
title: "Kaan Berke UGURLAR (@kaanberke) | Hashnode"
description: "Software Engineer. Read the latest articles by Kaan Berke UGURLAR on Hashnode."
url: "https://hashnode.com/@kaanberke"
---

@kaanberke

# Kaan Berke UGURLAR

@kaanberke · Joined November 2020

Software Engineer

### [The Ups and Downs of Probability Distributions in Deep Learning](https://kaanberke.hashnode.dev/the-ups-and-downs-of-probability-distributions-in-deep-learning?utm_source=hashnode&utm_medium=feed)

Understanding when probability distributions help or hinder deep learning model performance. Tips for effectively applying probability tools in NNs.

2 min read · [kaanberke.hashnode.dev](https://kaanberke.hashnode.dev/the-ups-and-downs-of-probability-distributions-in-deep-learning?utm_source=hashnode&utm_medium=feed)

[Nov 6, 2023](https://hashnode.com/posts/the-ups-and-downs-of-probability-distributions-in-deep-learning/654924d58d26a1e1dce2754d)

### [Linearity and Non-Linearity in Deep Learning: Why We Need Both](https://kaanberke.hashnode.dev/linearity-and-non-linearity-in-deep-learning-why-we-need-both?utm_source=hashnode&utm_medium=feed)

Linearity vs non-linearity in deep learning. How linear functions are limited and non-linear activations enable neural networks to model complex data.

3 min read · [kaanberke.hashnode.dev](https://kaanberke.hashnode.dev/linearity-and-non-linearity-in-deep-learning-why-we-need-both?utm_source=hashnode&utm_medium=feed)

[Oct 22, 2023](https://hashnode.com/posts/linearity-and-non-linearity-in-deep-learning-why-we-need-both/6535275725496c80f57be592)

### [Why Logarithmic Functions?](https://kaanberke.hashnode.dev/why-logarithmic-functions?utm_source=hashnode&utm_medium=feed)

Logarithms provide pivotal advantages for deep learning including numerical stability, probability measurement, and exponential linearization.

4 min read · [kaanberke.hashnode.dev](https://kaanberke.hashnode.dev/why-logarithmic-functions?utm_source=hashnode&utm_medium=feed)

[Oct 20, 2023](https://hashnode.com/posts/why-logarithmic-functions/6532aada18ec9f51a98fcf9a)

### [Mastering Batch Normalization for Neural Networks](https://kaanberke.hashnode.dev/mastering-batch-normalization-for-neural-networks?utm_source=hashnode&utm_medium=feed)

Batch normalization is an important technique for training neural networks. This guide explains batch normalization simply with an analogy, technical detail

2 min read · [kaanberke.hashnode.dev](https://kaanberke.hashnode.dev/mastering-batch-normalization-for-neural-networks?utm_source=hashnode&utm_medium=feed)

[Sep 23, 2023](https://hashnode.com/posts/mastering-batch-normalization-for-neural-networks/650f26d58c601bb9559add79)

### [This New Optimizer Called Lion Could Replace Adam as the Go-To for Training Neural Nets](https://kaanberke.hashnode.dev/this-new-optimizer-called-lion-could-replace-adam-as-the-go-to-for-training-neural-nets?utm_source=hashnode&utm_medium=feed)

Google Brain researchers discovered Lion, a new optimizer for training neural networks that outperforms Adam. Learn how Lion works and see results on vision

2 min read · [kaanberke.hashnode.dev](https://kaanberke.hashnode.dev/this-new-optimizer-called-lion-could-replace-adam-as-the-go-to-for-training-neural-nets?utm_source=hashnode&utm_medium=feed)

[Sep 17, 2023](https://hashnode.com/posts/this-new-optimizer-called-lion-could-replace-adam-as-the-go-to-for-training-neural-nets/650716cbe4e76b6f55414a46)
