Logo: to the web site of Uppsala University

Federated learning is a distributed and privacy-preserving approach to train a statistical model collaboratively from decentralized data held by different parties. However, when the datasets are not independent and identically distributed, models trained by naive federated algorithms may be biased towards certain participants, and model performance across participants is non-uniform. This is known as the fairness problem in federated learning. In this paper, we formulate fairness-controlled federated learning as a dynamical multi-objective optimization problem to ensure the fairness and convergence with theoretical guarantee. To solve the problem efficiently, we study the convergence and bias of Adam as the server optimizer in federated learning, and propose Adaptive Federated Adam ( AdaFedAdam ) to accelerate fair federated learning with alleviated bias. We validated the effectiveness, Pareto optimality and robustness of AdaFedAdam with numerical experiments and show that AdaFedAdam outperforms existing algorithms, providing better convergence and fairness properties of the federated scheme.

In many areas of data-driven science, large datasets are generated where the individual data objects are images, matrices, or otherwise have a clear structure. However, these objects can be information-sparse, and a challenge is to efficiently find and work with the most interesting data as early as possible in an analysis pipeline. We have recently proposed a new model for big data management where the internal structure and information of the data are associated with each data object (as opposed to simple metadata). There is then an opportunity for comprehensive data management solutions to account for data-specific internal structure as well as access patterns. In this article, we explore this idea together with our recently proposed hierarchical storage management framework that uses reinforcement learning (RL) for autonomous and dynamic data placement in different tiers in a storage hierarchy. Our case-study is based on four scientific datasets: Protein translocation microscopy images, Airfoil angle of attack meshes, 1000 Genomes sequences, and Phenotypic screening images. The presented results highlight that our framework is optimal and can quickly adapt to new data access requirements. It overall reduces the data processing time, and the proposed autonomous data placement is superior compared to any static or semi-static data placement policies.

Having access to sufficient data is essential in order to train accurate machine learning models, but much data is not publicly available. In drug discovery this is particularly evident, as much data is withheld at pharmaceutical companies for various reasons. Federated Learning (FL) aims at training a joint model between multiple parties but without disclosing data between the parties. In this work, we leverage Federated Learning to predict compound Mechanism of Action (MoA) using fluorescence image data from cell painting. Our study evaluates the effectiveness and efficiency of FL, comparing to non-collaborative and data-sharing collaborative learning in diverse scenarios. Specifically, we investigate the impact of data heterogeneity across participants on MoA prediction, an essential concern in real-life applications of FL, and demonstrate the benefits for all involved parties. This work highlights the potential of federated learning in multi-institutional collaborative machine learning for drug discovery and assessment of chemicals, offering a promising avenue to overcome data-sharing constraints.

Federated learning (FL) enables edge nodes to collaboratively contribute to constructing a global model without sharing their data. This is accomplished by devices computing local, private model updates that are then aggregated by a server. However, computational resource constraints and network communication can become a severe bottleneck for larger model sizes typical for deep learning (DL) applications. Edge nodes tend to have limited hardware resources (RAM, CPU), and the network bandwidth and reliability at the edge is a concern for scaling federated fleet applications. In this paper, we propose and evaluate a FL strategy inspired by transfer learning in order to reduce resource utilization on devices, as well as the load on the server and network in each global training round. For each local model update, we randomly select layers to train, freezing the remaining part of the model. In doing so, we can reduce both server load and communication costs per round by excluding all untrained layer weights from being transferred to the server. The goal of this study is to empirically explore the potential trade-off between resource utilization on devices and global model convergence under the proposed strategy. We implement the approach using the FL framework FEDn. A number of experiments were carried out over different datasets (CIFAR-10, CASA, and IMDB), performing different tasks using different DL model architectures. Our results show that training the model partially can accelerate the training process, efficiently utilizes resources on-device, and reduce the data transmission by around 75% and 53% when we train 25%, and 50% of the model layers, respectively, without harming the resulting global model accuracy. Furthermore, our results demonstrate a negative correlation between the number of participating clients in the training process and the number of layers that need to be trained on each client's side. As the number of clients increases, there is a decrease in the required number of layers. This observation highlights the potential of the approach, particularly in cross-device use cases.

During early development, cartilage provides shape and stability to the embryo while serving as a precursor for the skeleton. Correct formation of embryonic cartilage is hence essential for healthy development. In vertebrate cranial cartilage, it has been observed that a flat and laterally extended macroscopic geometry is linked to regular microscopic structure consisting of tightly packed, short, transversal clonar columns. However, it remains an ongoing challenge to identify how individual cells coordinate to successfully shape the tissue, and more precisely which mechanical interactions and cell behaviors contribute to the generation and maintenance of this columnar cartilage geometry during embryogenesis. Here, we apply a three-dimensional cell-based computational model to investigate mechanical principles contributing to column formation. The model accounts for clonal expansion, anisotropic proliferation and the geometrical arrangement of progenitor cells in space. We confirm that oriented cell divisions and repulsive mechanical interactions between cells are key drivers of column formation. In addition, the model suggests that column formation benefits from the spatial gaps created by the extracellular matrix in the initial configuration, and that column maintenance is facilitated by sequential proliferative phases. Our model thus correctly predicts the dependence of local order on division orientation and tissue thickness. The present study presents the first cell-based simulations of cell mechanics during cranial cartilage formation and we anticipate that it will be useful in future studies on the formation and growth of other cartilage geometries.

With the rapid development of big data and cloud computing, data management has become increasingly challenging. Over the years, a number of frameworks for data management have become available. Most of them are highly efficient, but ultimately create data silos. It becomes difficult to move and work coherently with data as new requirements emerge. A possible solution is to use an intelligent hierarchical (multi-tier) storage system (HSS). A HSS is a meta solution that consists of different storage frameworks organized as a jointly constructed storage pool. A built-in data migration policy that determines the optimal placement of the datasets in the hierarchy is essential. Placement decisions is a non-trivial task since it should be made according to the characteristics of the dataset, the tier status in a hierarchy, and access patterns. This paper presents an open-source hierarchical storage framework with a dynamic migration policy based on reinforcement learning (RL). We present a mathematical model, a software architecture, and implementations based on both simulations and a live cloud-based environment. We compare the proposed RL-based strategy to a baseline of three rule-based policies, showing that the RL-based policy achieves significantly higher efficiency and optimal data distribution in different scenarios.

With the rapid development of big data and cloud computing, data management has become increasingly challenging. A possible solution is to use an intelligent hierarchical (multi-tier) storage system (HSS). An HSS is a meta solution that consists of different storage frameworks organized as a jointly constructed storage pool. A built-in data migration policy that determines the optimal placement of the datasets in the hierarchy is essential. Placement decisions are a non-trivial task since they should be made according to the characteristics of the dataset, the tier status in a hierarchy, and access patterns. This paper presents an open-source hierarchical storage framework with a dynamic migration policy based on reinforcement learning (RL).

There are major differences in duration and scale at which limb development and regeneration proceed, raising the question to what extent regeneration is a recapitulation of development. We address this by analyzing skeletal elements using a combination of micro-CT imaging, molecular profiling and clonal cell tracing. We find that, in contrast to development, regenerative skeletal growth is accomplished based entirely on cartilage expansion prior to ossification, not limiting the transversal cartilage expansion and resulting in bulkier skeletal parts. The oriented extension of salamander cartilage and bone appear similar to the development of basicranial synchondroses in mammals, as we found no evidence for cartilage stem cell niches or growth plate-like structures during neither development nor regeneration. Both regenerative and developmental ossification in salamanders start from the cortical bone and proceeds inwards, showing the diversity of schemes for the synchrony of cortical and endochondral ossification among vertebrates. Normal limb development relies on synchronized formation of cartilage and bone. Here, the authors show that in salamander limb regeneration these processes are decoupled: ossification occurs after the final size of regenerating cartilage is reached, allowing fast regeneration and leading to bulky bones.

Background: Cell-based models are becoming increasingly popular for applications in developmental biology. However, the impact of numerical choices on the accuracy and efficiency of the simulation of these models is rarely meticulously tested. Without concrete studies to differentiate between solid model conclusions and numerical artifacts, modelers are at risk of being misled by their experiments’ results. Most cell-based modeling frameworks offer a feature-rich environment, providing a wide range of biological components, but are less suitable for numerical studies. There is thus a need for software specifically targeted at this use case.

Results: We present CBMOS, a Python framework for the simulation of the center-based or cell-centered model. Contrary to other implementations, CBMOS’ focus is on facilitating numerical study of center-based models by providing access to multiple ordinary differential equation solvers and force functions through a flexible, user-friendly interface and by enabling rapid testing through graphics processing unit (GPU) acceleration. We show-case its potential by illustrating two common workflows: (1) comparison of the numerical properties of two solvers within a Jupyter notebook and (2) measuring average wall times of both solvers on a high performance computing cluster. More specifically, we confirm that although for moderate accuracy levels the backward Euler method allows for larger time step sizes than the commonly used forward Euler method, its additional computational cost due to being an implicit method prohibits its use for practical test cases.

Conclusions: CBMOS is a flexible, easy-to-use Python implementation of the center-based model, exposing both basic model assumptions and numerical components to the user. It is available on GitHub and PyPI under an MIT license. CBMOS allows for fast prototyping on a central processing unit for small systems through the use of NumPy. Using CuPy on a GPU, cell populations of up to 10,000 cells can be simulated within a few seconds. As such, it will substantially lower the time investment for any modeler to check the crucial assumption that model conclusions are independent of numerical issues.