{
  "id": "2006.09637",
  "version": "v1",
  "published": "2020-06-17T03:45:25.000Z",
  "updated": "2020-06-17T03:45:25.000Z",
  "title": "FedCD: Improving Performance in non-IID Federated Learning",
  "authors": [
    "Kavya Kopparapu",
    "Eric Lin",
    "Jessica Zhao"
  ],
  "categories": [
    "cs.LG",
    "cs.DC",
    "stat.ML"
  ],
  "abstract": "Federated learning has been widely applied to enable decentralized devices, which each have their own local data, to learn a shared model. However, learning from real-world data can be challenging, as it is rarely identically and independently distributed (IID) across edge devices (a key assumption for current high-performing and low-bandwidth algorithms). We present a novel approach, FedCD, which clones and deletes models to dynamically group devices with similar data. Experiments on the CIFAR-10 dataset show that FedCD achieves higher accuracy and faster convergence compared to a FedAvg baseline on non-IID data while incurring minimal computation, communication, and storage overheads.",
  "revisions": [
    {
      "version": "v1",
      "updated": "2020-06-17T03:45:25.000Z"
    }
  ],
  "analyses": {
    "keywords": [
      "non-iid federated learning",
      "improving performance",
      "fedcd achieves higher accuracy",
      "edge devices",
      "low-bandwidth algorithms"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 0,
      "language": "en",
      "license": "arXiv",
      "status": "editable"
    }
  }
}