Adaptive Expert Models for Personalization in Federated Learning

Federated Learning (FL) is a promising framework for distributed learning when data is private and sensitive. However, the state-of-the-art solutions in this framework are not optimal when data is heterogeneous and non-Independent and Identically Distributed (non-IID). We propose a practical and robust approach to personalization in FL that adjusts to heterogeneous and non-IID data by balancing exploration and exploitation of several global models. To achieve our aim of personalization, we use a Mixture of Experts (MoE) that learns to group clients that are similar to each other, while using the global models more efficiently. We show that our approach achieves an accuracy up to 29.78 % and up to 4.38 % better compared to a local model in a pathological non-IID setting, even though we tune our approach in the IID setting.

~1 min read
Back to Top ↑


Adding Sparklines to LaTeX tables using Pandas

Tables in scientific papers often look less than professional, and sometimes this can even get in the way of understanding the message. In this blog post we will learn how to add sparklines to a LaTeX table, which not only makes your table stand out, but also allows for conveying information about for example trends in time-series.

4 min read

Create publication ready tables with Pandas

Tables in scientific papers often look less than professional, and sometimes this can even get in the way of understanding the message. In this blog post we will use pandas to automate making publication ready LaTeX tables that look great.

4 min read
Back to Top ↑


Bootstrapping your next LaTeX project

The process of setting up a new LaTeX project is made up of many manual steps, resulting in a patchwork that already from the start is not exercisable nor complete. In this post we will see how we can construct a solid starting point with a single command. This is part of a series to create the perfect open science git repository.

4 min read

LaTeX writing as a constrained non-convex optimization problem

The rejection rate for papers in good conferences is very high. To be accepted, a paper must not only be of a high scientific quality, but also at first impression perceived to be - or risk being thrown in the recycling bin. In this post we construct a system that automatically optimizes one proxy metric for perceived quality, removing one small frustrating step of scientific paper authorship and hopefully avoiding the bin.

13 min read

How to beat publisher PDF checks with LaTeX document unit testing

When submitting a scientific paper to a conference or a journal, there is often a mandatory step of passing the automated PDF checks set up by that publication. This step can often be nerve-racking and cause many hours of LaTeX troubleshooting. In this post we will create a series of test cases to catch these problems early in the writing process so that you can submit your manuscript only once.

14 min read

How to annoy your co-authors: a Gitlab CI pipeline for LaTeX

The process of writing a LaTeX document can be one full of manual steps, resulting in a patchwork document that is not exercisable nor complete. This makes it impossible to reproduce the document from code and data. In this post we will create a pipeline for compiling a LaTeX document that works both locally and using GitLab CI. This is part of a series to create the perfect open science git repository.

9 min read

Top 10 LaTeX packages for academic writing

In academic writing with LaTeX there are a lot of things that can be frustrating to the author. For many of these things there exists many packages that can help alleviate this frustration, but it is hard to find them. In this post I list 10 of my favorite packages to help remove some of this frustation, and make your papers look nicer so that you have a higher probability of getting your paper accepted. Hopefully.

4 min read

Reproducibility aspects of the Swedish COVID–19 estimate report

Researchers have called out for more transparency from The Public Health Agency of Sweden regarding the COVID-19 estimates for Sweden. Recently, a report has been released covering such estimates for the Stockholm region. Along the report, the code used for these estimates was uploaded to Github, which makes it possible for others to review and critique the work. In this post we will take a look at the reproducibility aspects of this release. We find that it is possible to some extent reproduce the figures in the report, and we suggest many improvements to the repository.

8 min read

Secure Federated Learning in 5G Mobile Networks

Machine Learning (ML) is an important enabler for optimizing, securing and managing mobile networks. This leads to increased collection and processing of data from network functions, which in turn may increase threats to sensitive end-user information. Consequently, mechanisms to reduce threats to end-user privacy are needed to take full advantage of ML. We seamlessly integrate Federated Learning (FL) into the 3GPP 5G Network Data Analytics (NWDA) architecture, and add a Multi-Party Computation (MPC) protocol for protecting the confidentiality of local updates. We evaluate the protocol and find that it has much lower overhead than previous work, without affecting ML performance.

~1 min read
Back to Top ↑


Create publication ready figures with Matplotlib and TikZ

Figures in scientific papers often look less than professional, and sometimes this can even get in the way of understanding the figure. In this blog post we show how to use matplotlib and tikzplotlib to make publication ready figures that look great and can be styled from the document preamble. Beautiful and understandable figures can possibly lead to higher publication acceptance rate, at least I hope so…

6 min read
Back to Top ↑


He makes intelligent systems even smarter

In the future, different intelligent systems will need to share data and experiences with each other to become good enough for certain tasks. RISE Industrial PhD student Martin Isaksson’s research is an important step on the way. This area is highly relevant for Ericsson and the development of its 5G but lessons learned along the way might open for many more solutions.

~1 min read

Predicting strongest cell on secondary carrier using primary carrier data

There is a rapid evolution in telecommunication with denser networks and systems operating on an increasing number of frequency bands. Also in the next generation 5G networks, even further densification is needed to be able to reach the tight requirements, implying more nodes on each carrier. Denser networks and more frequencies makes it challenging to ensure the best possible cell and frequency carrier assignment to a User Equipment (UE), without the UE needing to perform an excessive amount of inter-frequency measurements and reporting. In this paper, we propose a procedure of predicting the strongest cell of a secondary carrier, and the procedure is exemplified in a UE load-balancing use case. The prediction is based on only measurements on the primary carrier cells, avoiding costly inter-frequency measurements. Simulations of a realistic network deployment show that a UE selection based on the proposed secondary carrier prediction is significantly better than a random UE selection for load balancing.

~1 min read
Back to Top ↑


The F Pie Initiative

Sometimes, when I come across a Powerpoint presentation with multiple multi-slice pie charts on a single slide my head hurts and I have to go air-poop for half-an-hour. Now I have decided to take action!

1 min read

Washing apparatus ready detection (WARD)

Using a PocketLab put on the top of a washing machine, we measure the magnetic field generated by the motor. The magnitude of this vector is used to detect if the motor is on, and when the motor has been off for some time, we say that the cycle is finished.

1 min read
Back to Top ↑