---
title: "Stephane Bersier (@Bersier) | Hashnode"
description: "Stephane Bersier (@Bersier) writes on Hashnode. Read their latest articles and follow their work."
url: "https://hashnode.com/@Bersier"
---

@Bersier

# Stephane Bersier

@Bersier · Joined December 2022

[Website](https://bersier.github.io/)

### [Complex Log-Mean-Exp Networks](https://aiconversations.hashnode.dev/complex-log-mean-exp-networks?utm_source=hashnode&utm_medium=feed)

1\. Core definition A complex log-mean-exp unit (LME unit) is defined as $$y = \\frac{1}{\\beta}\\,\\log\\Big(\\widetilde{\\sum\_{i=1}^n w\_i \\exp (\\beta \\, x\_i)} \\Big)$$where \\(x\_i \\in \\mathbb{C}\\) are the complex inputs. \\(y\\) is the unit’s output. \\(w\_i ...

3 min read · [aiconversations.hashnode.dev](https://aiconversations.hashnode.dev/complex-log-mean-exp-networks?utm_source=hashnode&utm_medium=feed)

[Oct 29, 2025](https://hashnode.com/posts/complex-log-mean-exp-networks/69028431e04ca2c77b54ab93)

### [Less Overfitting via Stochastic Exposure](https://aiconversations.hashnode.dev/less-overfitting-via-stochastic-exposure?utm_source=hashnode&utm_medium=feed)

Using stochastic exposure for neural network calibration and reduced overfitting

5 min read · [aiconversations.hashnode.dev](https://aiconversations.hashnode.dev/less-overfitting-via-stochastic-exposure?utm_source=hashnode&utm_medium=feed)

[Sep 8, 2025](https://hashnode.com/posts/less-overfitting-via-stochastic-exposure/68bef81a566a31f99d3965bf)

### [Why Transformers Are Powerful](https://aiconversations.hashnode.dev/why-transformers-are-powerful?utm_source=hashnode&utm_medium=feed)

The following is an edited version of a synthesis written by ChatGPT. 1. Attention as Dynamic Selectivity The attention mechanism enables transformers to dynamically route information. Rather than collapsing inputs into a fixed summary, each token se...

2 min read · [aiconversations.hashnode.dev](https://aiconversations.hashnode.dev/why-transformers-are-powerful?utm_source=hashnode&utm_medium=feed)

[Aug 31, 2025](https://hashnode.com/posts/why-transformers-are-powerful/68b4c2fffd77b62d03276fc2)

### [On the Power of Attention](https://aiconversations.hashnode.dev/on-the-power-of-attention?utm_source=hashnode&utm_medium=feed)

The following is a synthesis written by ChatGPT. Attention as Selective Focus The central problem is that a processing system (whether a brain’s conscious workspace or a machine learning model) has finite capacity. It cannot load all available inform...

3 min read · [aiconversations.hashnode.dev](https://aiconversations.hashnode.dev/on-the-power-of-attention?utm_source=hashnode&utm_medium=feed)

[Aug 31, 2025](https://hashnode.com/posts/on-the-power-of-attention/68b4bf969022fe2b543518f1)

### [NN Architectures as Generalized Algorithms](https://aiconversations.hashnode.dev/nn-architectures-as-generalized-algorithms?utm_source=hashnode&utm_medium=feed)

The following is an edited version of a synthesis written by ChatGPT. 1) Core thesis We can treat neural architectures as a generalization of algorithms. There is a spectrum of algorithmic content: On one end, simple feedforward MLPs carry almost ...

4 min read · [aiconversations.hashnode.dev](https://aiconversations.hashnode.dev/nn-architectures-as-generalized-algorithms?utm_source=hashnode&utm_medium=feed)

[Aug 18, 2025](https://hashnode.com/posts/nn-architectures-as-generalized-algorithms/68a33cce4b7a2f9bf1157725)
