Federated learning (FL) is a machine learning setting where many clients (e.g. mobile devices or whole organizations) collaboratively train a model under the orchestration of a central server (e.g. service provider), while keeping the training data decentralized. FL embodies the principles of focused data collection and minimization, and can mitigate many of the systemic privacy risks and costs resulting from traditional, centralized machine learning and data science approaches. Motivated by the explosive growth in FL research, this paper discusses recent advances and presents an extensive collection of open problems and challenges.

Advances and Open Problems in Federated Learning

In recent years, mobile devices are equipped with increasingly advanced sensing and computing capabilities. Coupled with advancements in Deep Learning (DL), this opens up countless possibilities for meaningful applications, e.g., for medical purposes and in vehicular networks. Traditional cloud-based Machine Learning (ML) approaches require the data to be centralized in a cloud server or data center. However, this results in critical issues related to unacceptable latency and communication inefficiency. To this end, Mobile Edge Computing (MEC) has been proposed to bring intelligence closer to the edge, where data is produced. However, conventional enabling technologies for ML at mobile edge networks still require personal data to be shared with external parties, e.g., edge servers. Recently, in light of increasingly stringent data privacy legislations and growing privacy concerns, the concept of Federated Learning (FL) has been introduced. In FL, end devices use their local data to train an ML model required by the server. The end devices then send the model updates rather than raw data to the server for aggregation. FL can serve as an enabling technology in mobile edge networks since it enables the collaborative training of an ML model and also enables DL for mobile edge network optimization. However, in a large-scale and complex mobile edge network, heterogeneous devices with varying constraints are involved. This raises challenges of communication costs, resource allocation, and privacy and security in the implementation of FL at scale. In this survey, we begin with an introduction to the background and fundamentals of FL. Then, we highlight the aforementioned challenges of FL implementation and review existing solutions. Furthermore, we present the applications of FL for mobile edge network optimization. Finally, we discuss the important challenges and future research directions in FL.

/pdf/federated-learning-in-mobile-edge-networks-a-comprehensive-5dgelc5m6h.pdf

Federated Learning in Mobile Edge Networks: A Comprehensive Survey

In recent years, mobile devices are equipped with increasingly advanced sensing and computing capabilities. Coupled with advancements in Deep Learning (DL), this opens up countless possibilities for meaningful applications. Traditional cloudbased Machine Learning (ML) approaches require the data to be centralized in a cloud server or data center. However, this results in critical issues related to unacceptable latency and communication inefficiency. To this end, Mobile Edge Computing (MEC) has been proposed to bring intelligence closer to the edge, where data is produced. However, conventional enabling technologies for ML at mobile edge networks still require personal data to be shared with external parties, e.g., edge servers. Recently, in light of increasingly stringent data privacy legislations and growing privacy concerns, the concept of Federated Learning (FL) has been introduced. In FL, end devices use their local data to train an ML model required by the server. The end devices then send the model updates rather than raw data to the server for aggregation. FL can serve as an enabling technology in mobile edge networks since it enables the collaborative training of an ML model and also enables DL for mobile edge network optimization. However, in a large-scale and complex mobile edge network, heterogeneous devices with varying constraints are involved. This raises challenges of communication costs, resource allocation, and privacy and security in the implementation of FL at scale. In this survey, we begin with an introduction to the background and fundamentals of FL. Then, we highlight the aforementioned challenges of FL implementation and review existing solutions. Furthermore, we present the applications of FL for mobile edge network optimization. Finally, we discuss the important challenges and future research directions in FL

Federated learning is a distributed machine learning paradigm in which a large number of clients coordinate with a central server to learn a model without sharing their own training data. Standard federated optimization methods such as Federated Averaging (FedAvg) are often difficult to tune and exhibit unfavorable convergence behavior. In non-federated settings, adaptive optimization methods have had notable success in combating such issues. In this work, we propose federated versions of adaptive optimizers, including Adagrad, Adam, and Yogi, and analyze their convergence in the presence of heterogeneous data for general non-convex settings. Our results highlight the interplay between client heterogeneity and communication efficiency. We also perform extensive experiments on these methods and show that the use of adaptive optimizers can significantly improve the performance of federated learning.

Adaptive Federated Optimization

Distribution shifts -- where the training distribution differs from the test distribution -- can substantially degrade the accuracy of machine learning (ML) systems deployed in the wild. Despite their ubiquity, these real-world distribution shifts are under-represented in the datasets widely used in the ML community today. To address this gap, we present WILDS, a curated collection of 8 benchmark datasets that reflect a diverse range of distribution shifts which naturally arise in real-world applications, such as shifts across hospitals for tumor identification; across camera traps for wildlife monitoring; and across time and location in satellite imaging and poverty mapping. On each dataset, we show that standard training results in substantially lower out-of-distribution than in-distribution performance, and that this gap remains even with models trained by existing methods for handling distribution shifts. This underscores the need for new training methods that produce models which are more robust to the types of distribution shifts that arise in practice. To facilitate method development, we provide an open-source package that automates dataset loading, contains default model architectures and hyperparameters, and standardizes evaluations. Code and leaderboards are available at this https URL.

WILDS: A Benchmark of in-the-Wild Distribution Shifts

Modern federated networks, such as those comprised of wearable devices, mobile phones, or autonomous vehicles, generate massive amounts of data each day. This wealth of data can help to learn models that can improve the user experience on each device. However, the scale and heterogeneity of federated data presents new challenges in research areas such as federated learning, meta-learning, and multi-task learning. As the machine learning community begins to tackle these challenges, we are at a critical time to ensure that developments made in these areas are grounded with realistic benchmarks. To this end, we propose LEAF, a modular benchmarking framework for learning in federated settings. LEAF includes a suite of open-source federated datasets, a rigorous evaluation framework, and a set of reference implementations, all geared towards capturing the obstacles and intricacies of practical federated environments.

LEAF: A Benchmark for Federated Settings

OBJECTIVES The management of dog bite wounds is controversial, and current data on risk of infection are variable and inconsistent. Furthermore, the use of prophylactic or empiric antibiotics for the treatment of these wounds is debatable. We investigate the rate of wound infections and other complications after primary repair of pediatric facial dog bite injuries. METHODS We reviewed 87 consecutive patients aged 18 years or younger who had facial dog bite injuries from January 2003 to December 2008. Variables examined were age, sex, setting of repair, number of sutures used for repair, whether surgical drains were used, and antibiotic administration. End points measured were incidence of wound infection, need for scar revision, and any wound complications. RESULTS The mean age of patients was 6.8 years, and the majority were women (53%). All facial injuries were primarily repaired at the time of presentation either in the emergency department (ED; 46%), operating room (OR; 51%), or an outpatient setting (3%). All patients received an antibiotic course, none of the patients developed wound infection, and no subsequent scar revisions were performed. Three patients repaired in the OR underwent placement of a total of 4 closed-suction drains. The mean (SD) age of patients repaired in the OR was significantly younger than those repaired in the ED (5.7 [3.9] vs 8.0 [4.5] years, respectively; P < 0.01). The number of sutures used were greater for patients repaired in the OR than in the ED (66.4 [39.6] vs 21.7 [12.5], respectively; P < 0.01). CONCLUSIONS Intuitively, younger patients and patients with greater severity injuries are more likely to undergo repair in the OR, and this was supported by our data. Overall, we found that primary repair of pediatric facial dog bite injuries, including complex soft-tissue injuries, is safe when performed in conjunction with antibiotic administration; however, further cross-specialty studies are needed to fully characterize these end points in a larger population.

Primary repair of facial dog bite injuries in children.

The natural world is abundant with concepts expressed via visual, acoustic, tactile, and linguistic modalities. Much of the existing progress in multimodal learning, however, focuses primarily on problems where the same set of modalities are present at train and test time, which makes learning in low-resource modalities particularly difficult. In this work, we propose algorithms for cross-modal generalization: a learning paradigm to train a model that can (1) quickly perform new tasks in a target modality (i.e. meta-learning) and (2) doing so while being trained on a different source modality. We study a key research question: how can we ensure generalization across modalities despite using separate encoders for different source and target modalities? Our solution is based on meta-alignment, a novel method to align representation spaces using strongly and weakly paired cross-modal data while ensuring quick generalization to new tasks across different modalities. We study this problem on 3 classification tasks: text to image, image to audio, and text to speech. Our results demonstrate strong performance even when the new target modality has only a few (1-10) labeled samples and in the presence of noisy labels, a scenario particularly prevalent in low-resource modalities.

Cross-Modal Generalization: Learning in Low Resource Modalities via Meta-Alignment

Objectives: Extensive perineal resections often require autologous tissue reconstruction, especially in wide oncological resections. Local and regional pedicled flaps from the lower extremity and abdominal sites have been described. Defects of the pelvis and perineum rarely require free-tissue transfer. The vertical rectus abdominis myocutaneous (VRAM) flap, traditionally delivered to the perineum through an intraperitoneal transpelvic route, is a workhorse flap for combined pelvic and perineal defects because of its ability to provide substantial coverage of the perineum, reliable vascular supply, and larger volume to obliterate dead space. We propose and describe an extended VRAM flap for vulvar reconstruction delivered to the perineum in an extrapelvic fashion. Methods: A 54-year-old woman with a prior history of anal squamous cell carcinoma underwent neoadjuvant chemoradiotherapy followed by abdominoperineal resection, total abdominal hysterectomy, and bilateral salpingo-oophorectomy. Three years later, she developed vulvar squamous cell carcinoma with vascular and lymphatic invasion and underwent radical vulvectomy and distal urethrectomy. The resection defect was 10 × 15 cm, including the distal 1 cm of the urethra, distal vaginal orifice, and wide exposure of the pubic bone. An extrapelvic extended VRAM flap was used for reconstruction. Results: The flap was harvested and transposed into the defect via a wide suprapubic subcutaneous tunnel. A neovaginal and urethral orifice was created in the flap by splitting the muscle in the direction of its fibers, taking care to protect the vascular pedicle, and inset to the vaginal orifice. There were no postoperative complications. She has maintained urinary continence with follow-up of 38 months. Conclusions: Introduction of a rectus abdominis flap to the perineum through an extrapelvic route is preferred if laparotomy is not used for the resection. We successfully report and advocate the use of an extended VRAM flap for vulvar reconstruction delivered to the perineum in an extrapelvic fashion.

Perineal reconstruction with an extrapelvic vertical rectus abdominis myocutaneous flap.

In the articulatory synthesis task, speech is synthesized from input features containing information about the physical behavior of the human vocal tract. This task provides a promising direction for speech synthesis research, as the articulatory space is compact, smooth, and interpretable. Current works have highlighted the potential for deep learning models to perform articulatory synthesis. However, it remains unclear whether these models can achieve the efficiency and fidelity of the human speech production system. To help bridge this gap, we propose a time-domain articulatory synthesis methodology and demonstrate its efficacy with both electromagnetic articulography (EMA) and synthetic articulatory feature inputs. Our model is computationally efficient and achieves a transcription word error rate (WER) of 18.5% for the EMA-to-speech task, yielding an improvement of 11.6% compared to prior work. Through interpolation experiments, we also highlight the generalizability and interpretability of our approach.

/pdf/deep-speech-synthesis-from-articulatory-representations-3t6qq7ke.pdf

Peter Wu

Papers

LEAF: A Benchmark for Federated Settings

Primary repair of facial dog bite injuries in children.

Cross-Modal Generalization: Learning in Low Resource Modalities via Meta-Alignment

Perineal reconstruction with an extrapelvic vertical rectus abdominis myocutaneous flap.

Deep Speech Synthesis from Articulatory Representations