Federated learning (also known as collaborative learning) is a sub-field of machine learning focusing on settings in which multiple entities (often referred to as clients) collaboratively train a model while ensuring that their data remains decentralized.[1] This stands in contrast to machine learning settings in which data is centrally stored. One of the primary defining characteristics of federated learning is data heterogeneity. Due to the decentralized nature of the clients' data, there is no guarantee that data samples held by each client are independently and identically distributed.
Federated learning is generally concerned with and motivated by issues such as data privacy, data minimization, and data access rights. Its applications involve a variety of research areas including defense, telecommunications, the Internet of Things, and pharmaceuticals.
Federated learning aims at training a machine learning algorithm, for instance deep neural networks, on multiple local datasets contained in local nodes without explicitly exchanging data samples. The general principle consists in training local models on local data samples and exchanging parameters (e.g. the weights and biases of a deep neural network) between these local nodes at some frequency to generate a global model shared by all nodes.
The main difference between federated learning and distributed learning lies in the assumptions made on the properties of the local datasets,[2] as distributed learning originally aims at parallelizing computing power where federated learning originally aims at training on heterogeneous datasets. While distributed learning also aims at training a single model on multiple servers, a common underlying assumption is that the local datasets are independent and identically distributed (i.i.d.) and roughly have the same size. None of these hypotheses are made for federated learning; instead, the datasets are typically heterogeneous and their sizes may span several orders of magnitude. Moreover, the clients involved in federated learning may be unreliable as they are subject to more failures or drop out since they commonly rely on less powerful communication media (i.e. Wi-Fi) and battery-powered systems (i.e. smartphones and IoT devices) compared to distributed learning where nodes are typically datacenters that have powerful computational capabilities and are connected to one another with fast networks.[3]
The objective function for federated learning is as follows:
f(x1,...,xK)=\dfrac{1}{K}
| K | |
| \sum | |
| i=1 |
fi(xi)
where
K
xi
i
fi
i
xi
i
The goal of federated learning is to train a common model on all of the nodes' local datasets, in other words:
f(x1,...,xK)
xi
x1,...,xK
x
In the centralized federated learning setting, a central server is used to orchestrate the different steps of the algorithms and coordinate all the participating nodes during the learning process. The server is responsible for the nodes selection at the beginning of the training process and for the aggregation of the received model updates. Since all the selected nodes have to send updates to a single entity, the server may become a bottleneck of the system.
In the decentralized federated learning setting, the nodes are able to coordinate themselves to obtain the global model. This setup prevents single point failures as the model updates are exchanged only between interconnected nodes without the orchestration of the central server. Nevertheless, the specific network topology may affect the performances of the learning process. See blockchain-based federated learning[4] and the references therein.
An increasing number of application domains involve a large set of heterogeneous clients, e.g., mobile phones and IoT devices.[5] Most of the existing Federated learning strategies assume that local models share the same global model architecture. Recently, a new federated learning framework named HeteroFL was developed to address heterogeneous clients equipped with very different computation and communication capabilities.[6] The HeteroFL technique can enable the training of heterogeneous local models with dynamically varying computation and non-IID data complexities while still producing a single accurate global inference model.[6] [7]
To ensure good task performance of a final, central machine learning model, federated learning relies on an iterative process broken up into an atomic set of client-server interactions known as a federated learning round. Each round of this process consists in transmitting the current global model state to participating nodes, training local models on these local nodes to produce a set of potential model updates at each node, and then aggregating and processing these local updates into a single global update and applying it to the global model.
In the methodology below, a central server is used for aggregation, while local nodes perform local training depending on the central server's orders. However, other strategies lead to the same results without central servers, in a peer-to-peer approach, using gossip[8] or consensus methodologies.[9]
Assuming a federated round composed by one iteration of the learning process, the learning procedure can be summarized as follows:[10]
The procedure considered before assumes synchronized model updates. Recent federated learning developments introduced novel techniques to tackle asynchronicity during the training process, or training with dynamically varying models.[6] Compared to synchronous approaches where local models are exchanged once the computations have been performed for all layers of the neural network, asynchronous ones leverage the properties of neural networks to exchange model updates as soon as the computations of a certain layer are available. These techniques are also commonly referred to as split learning[11] [12] and they can be applied both at training and inference time regardless of centralized or decentralized federated learning settings.[6]
In most cases, the assumption of independent and identically distributed samples across local nodes does not hold for federated learning setups. Under this setting, the performances of the training process may vary significantly according to the unbalanced local data samples as well as the particular probability distribution of the training examples (i.e., features and labels) stored at the local nodes. To further investigate the effects of non-IID data, the following description considers the main categories presented in the preprint by Peter Kairouz et al. from 2019.
The description of non-IID data relies on the analysis of the joint probability between features and labels for each node.This allows decoupling of each contribution according to the specific distribution available at the local nodes.The main categories for non-iid data can be summarized as follows:
The loss in accuracy due to non-iid data can be bounded through using more sophisticated means of doing data normalization, rather than batch normalization.[13]
The way the statistical local outputs are pooled and the way the nodes communicate with each other can change from the centralized model explained in the previous section. This leads to a variety of federated learning approaches: for instance no central orchestrating server, or stochastic communication.[14]
In particular, orchestrator-less distributed networks are one important variation. In this case, there is no central server dispatching queries to local nodes and aggregating local models. Each local node sends its outputs to several randomly-selected others, which aggregate their results locally. This restrains the number of transactions, thereby sometimes reducing training time and computing cost.[15]
Once the topology of the node network is chosen, one can control different parameters of the federated learning process (in addition to the machine learning model's own hyperparameters) to optimize learning:
T
K
C
B
Other model-dependent parameters can also be tinkered with, such as:
N
η
Those parameters have to be optimized depending on the constraints of the machine learning application (e.g., available computing power, available memory, bandwidth). For instance, stochastically choosing a limited fraction
C
Federated learning requires frequent communication between nodes during the learning process. Thus, it requires not only enough local computing power and memory, but also high bandwidth connections to be able to exchange parameters of the machine learning model. However, the technology also avoids data communication, which can require significant resources before starting centralized machine learning. Nevertheless, the devices typically employed in federated learning are communication-constrained, for example IoT devices or smartphones are generally connected to Wi-Fi networks, thus, even if the models are commonly less expensive to be transmitted compared to raw data, federated learning mechanisms may not be suitable in their general form.
Federated learning raises several statistical challenges:
A number of different algorithms for federated optimization have been proposed.
Deep learning training mainly relies on variants of stochastic gradient descent, where gradients are computed on a random subset of the total dataset and then used to make one step of the gradient descent.
Federated stochastic gradient descent[19] is the direct transposition of this algorithm to the federated setting, but by using a random fraction
C
Federated averaging (FedAvg) is a generalization of FedSGD, which allows local nodes to perform more than one batch update on local data and exchanges the updated weights rather than the gradients. The rationale behind this generalization is that in FedSGD, if all local nodes start from the same initialization, averaging the gradients is strictly equivalent to averaging the weights themselves. Further, averaging tuned weights coming from the same initialization does not necessarily hurt the resulting averaged model's performance.[20] Variations of FedAvg based on adaptive optimizers such as ADAM and AdaGrad have been proposed, and generally outperform FedAvg.
Federated learning methods suffer when the device datasets are heterogeneously distributed. Fundamental dilemma in heterogeneously distributed device setting is that minimizing the device loss functions is not the same as minimizing the global loss objective. In 2021, Acar et al.[21] introduced FedDyn method as a solution to heterogenous dataset setting. FedDyn dynamically regularizes each devices loss function so that the modified device losses converges to the actual global loss. Since the local losses are aligned, FedDyn is robust to the different heterogeneity levels and it can safely perform full minimization in each device. Theoretically, FedDyn converges to the optimal (a stationary point for nonconvex losses) by being agnostic to the heterogeneity levels. These claims are verified with extensive experimentations on various datasets.
Minimizing the number of communications is the gold-standard for comparison in federated learning. We may also want to decrease the local computation levels per device in each round. FedDynOneGD is an extension of FedDyn with less local compute requirements. FedDynOneGD calculates only one gradients per device in each round and update the model with a regularized version of the gradient. Hence, the computation complexity is linear in local dataset size. Moreover, gradient computation can be parallelizable within each device which is different from successive SGD steps. Theoretically, FedDynOneGD achieves the same convergence guarantees as in FedDyn with less local computation.
Federated Learning methods cannot achieve good global performance under non-IID settings which motivates the participating clients to yield personalized models in federation. Recently, Vahidian et al.[22] introduced Sub-FedAvg opening a new personalized FL algorithm paradigm by proposing Hybrid Pruning (structured + unstructured pruning) with averaging on the intersection of clients’ drawn subnetworks which simultaneously handles communication efficiency, resource constraints and personalized models accuracies.
Sub-FedAvg is the first work which shows existence of personalized winning tickets for clients in federated learning through experiments. Moreover, it also proposes two algorithms on how to effectively draw the personalized subnetworks. Sub-FedAvg tries to extend the "lottery ticket hypothesis" which is for centrally trained neural networks to federated learning trained neural networks leading to this open research problem: “Do winning tickets exist for clients’ neural networks being trained in federated learning? If yes, how to effectively draw the personalized subnetworks for each client?”
IDA (Inverse Distance Aggregation) is a novel adaptive weighting approach for clients based on meta-information which handles unbalanced and non-iid data. It uses the distance of the model parameters as a strategy to minimize the effect of outliers and improve the model's convergence rate.[23]
Very few methods for hybrid federated learning, where clients only hold subsets of both features and samples, exist. Yet, this scenario is very important in practical settings. Hybrid Federated Dual Coordinate Ascent (HyFDCA)[24] is a novel algorithm proposed in 2024 that solves convex problems in the hybrid FL setting. This algorithm extends CoCoA, a primal-dual distributed optimization algorithm introduced by Jaggi et al. (2014)[25] and Smith et al. (2017),[26] to the case where both samples and features are partitioned across clients.
HyFDCA claims several improvement over existing algorithms:
There is only one other algorithm that focuses on hybrid FL, HyFEM proposed by Zhang et al. (2020).[29] This algorithm uses a feature matching formulation that balances clients building accurate local models and the server learning an accurate global model. This requires a matching regularizer constant that must be tuned based on user goals and results in disparate local and global models. Furthermore, the convergence results provided for HyFEM only prove convergence of the matching formulation not of the original global problem. This work is substantially different than HyFDCA's approach which uses data on local clients to build a global model that converges to the same solution as if the model was trained centrally. Furthermore, the local and global models are synchronized and do not require the adjustment of a matching parameter between local and global models. However, HyFEM is suitable for a vast array of architectures including deep learning architectures, whereas HyFDCA is designed for convex problems like logistic regression and support vector machines.
HyFDCA is empirically benchmarked against the aforementioned HyFEM as well as the popular FedAvg in solving convex problem (specifically classification problems) for several popular datasets (MNIST, Covtype, and News20). The authors found HyFDCA converges to a lower loss value and higher validation accuracy in less overall time in 33 of 36 comparisons examined and 36 of 36 comparisons examined with respect to the number of outer iterations.[24] Lastly, HyFDCA only requires tuning of one hyperparameter, the number of inner iterations, as opposed to FedAvg (which requires tuning three) or HyFEM (which requires tuning four). In addition to FedAvg and HyFEM being quite difficult to optimize hyperparameters in turn greatly affecting convergence, HyFDCA's single hyperparameter allows for simpler practical implementations and hyperparameter selection methodologies.
Federated Learning (FL) provides training of global shared model using decentralized data sources on edge nodes while preserving data privacy. However, its performance in the computer vision applications using Convolution neural network (CNN) considerably behind that of centralized training due to limited communication resources and low processing capability at edge nodes. Alternatively, Pure Vision transformer models (VIT) outperform CNNs by almost four times when it comes to computational efficiency and accuracy. Hence, we propose a new FL model with reconstructive strategy called FED-REV, Illustrates how attention-based structures (pure Vision Transformers) enhance FL accuracy over large and diverse data distributed over edge nodes, in addition to the proposed reconstruction strategy that determines the dimensions influence of each stage of the vision transformer and then reduce its dimension complexity which reduce computation cost of edge devices in addition to preserving accuracy achieved due to using the pure Vision transformer.[30]
Federated learning has started to emerge as an important research topic in 2015 and 2016,[31] with the first publications on federated averaging in telecommunication settings. Before that, in a thesis work titled "A Framework for Multi-source Prefetching Through Adaptive Weight",[32] an approach to aggregate predictions from multiple models trained at three location of a request response cycle with was proposed. Another important aspect of active research is the reduction of the communication burden during the federated learning process. In 2017 and 2018, publications have emphasized the development of resource allocation strategies, especially to reduce communication requirements[33] between nodes with gossip algorithms[34] as well as on the characterization of the robustness to differential privacy attacks.[35] Other research activities focus on the reduction of the bandwidth during training through sparsification and quantization methods,[33] where the machine learning models are sparsified and/or compressed before they are shared with other nodes. Developing ultra-light DNN architectures is essential for device-/edge- learning and recent work recognises both the energy efficiency requirements [36] for future federated learning and the need to compress deep learning, especially during learning.[37]
Recent research advancements are starting to consider real-world propagating channels[38] as in previous implementations ideal channels were assumed. Another active direction of research is to develop Federated learning for training heterogeneous local models with varying computation complexities and producing a single powerful global inference model.[6]
A learning framework named Assisted learning was recently developed to improve each agent's learning capabilities without transmitting private data, models, and even learning objectives.[39] Compared with Federated learning that often requires a central controller to orchestrate the learning and optimization, Assisted learning aims to provide protocols for the agents to optimize and learn among themselves without a global model.
Federated learning typically applies when individual actors need to train models on larger datasets than their own, but cannot afford to share the data in itself with others (e.g., for legal, strategic or economic reasons). The technology yet requires good connections between local servers and minimum computational power for each node.
Self-driving cars encapsulate many machine learning technologies to function: computer vision for analyzing obstacles, machine learning for adapting their pace to the environment (e.g., bumpiness of the road). Due to the potential high number of self-driving cars and the need for them to quickly respond to real world situations, traditional cloud approach may generate safety risks. Federated learning can represent a solution for limiting volume of data transfer and accelerating learning processes.[40] [41]
In Industry 4.0, there is a widespread adoption of machine learning techniques[42] to improve the efficiency and effectiveness of industrial process while guaranteeing a high level of safety. Nevertheless, privacy of sensitive data for industries and manufacturing companies is of paramount importance. Federated learning algorithms can be applied to these problems as they do not disclose any sensitive data. In addition, FL also implemented for PM2.5 prediction to support Smart city sensing applications.[43]
Federated learning seeks to address the problem of data governance and privacy by training algorithms collaboratively without exchanging the data itself. Today's standard approach of centralizing data from multiple centers comes at the cost of critical concerns regarding patient privacy and data protection. To solve this problem, the ability to train machine learning models at scale across multiple medical institutions without moving the data is a critical technology. Nature Digital Medicine published the paper "The Future of Digital Health with Federated Learning"[44] in September 2020, in which the authors explore how federated learning may provide a solution for the future of digital health, and highlight the challenges and considerations that need to be addressed. Recently, a collaboration of 20 different institutions around the world validated the utility of training AI models using federated learning. In a paper published in Nature Medicine "Federated learning for predicting clinical outcomes in patients with COVID-19",[45] they showcased the accuracy and generalizability of a federated AI model for the prediction of oxygen needs in patients with COVID-19 infections. Furthermore, in a published paper "A Systematic Review of Federated Learning in the Healthcare Area: From the Perspective of Data Properties and Applications", the authors trying to provide a set of challenges on FL challenges on medical data-centric perspective.[46]
A coalition from industry and academia has developed MedPerf,[47] an open source platform that enables validation of medical AI models in real world data. The platform relies technically on federated evaluation of AI models aiming to alleviate concerns of patient privacy and conceptually on diverse benchmark committees to build the specifications of neutral clinically impactful benchmarks.[48]
Robotics includes a wide range of applications of machine learning methods: from perception and decision-making to control. As robotic technologies have been increasingly deployed from simple and repetitive tasks (e.g. repetitive manipulation) to complex and unpredictable tasks (e.g. autonomous navigation), the need for machine learning grows. Federated Learning provides a solution to improve over conventional machine learning training methods. In the paper,[49] mobile robots learned navigation over diverse environments using the FL-based method, helping generalization. In the paper,[50] Federated Learning is applied to improve multi-robot navigation under limited communication bandwidth scenarios, which is a current challenge in real-world learning-based robotic tasks. In the paper,[51] Federated Learning is used to learn Vision-based navigation, helping better sim-to-real transfer.