Open Problems

Federated Learning

Eliminating Server-Side Auxiliary Data Dependencies in Federated Model Alignment and Calibration

Barrier to removePartly addressed
Possible candidate · 3/5 runs4 papers report this0% from 2025+

Generated automatically from the limitations stated in 4 papers (NeurIPS, ICML, ICCV), listed under Evidence. It is not a paper, and it does not come from papers submitted to CSPaper.

The problem

Many advanced federated learning (FL) algorithms for feature alignment, knowledge distillation, and client temperature calibration require an auxiliary (public or proxy) dataset located at the central server. In privacy-critical cross-silo or cross-device deployments (such as medical imaging or sensitive mobile data), representative public proxy data simply does not exist or cannot be stored on the server due to compliance constraints. Consequently, these federated methods cannot be deployed in true zero-server-data environments, forcing practitioners to fall back to basic parameter-averaging baselines that suffer severe degradation under heterogeneous data distributions.

Why it matters

Enables advanced federated distillation, alignment, and adaptive calibration techniques to run in strictly private FL pipelines where zero central data is accessible.

Ways to approach it

Prior-work checks are free with an account. Results someone already ran are shown to everyone.

  1. 1

    Data-Free Synthetic Feature Synthesis: Develop server-side generative inversion techniques that reconstruct class-conditional pseudo-representations directly from shared client classifier statistics or low-dimensional weight matrices, measuring alignment accuracy and communication overhead compared to real proxy datasets on heterogeneous CIFAR-100 and FLAIR benchmarks.

  2. 2

    Federated Decentralized Calibration: Reformulate server-side temperature and alignment estimation as a privacy-preserving consensus protocol executed across local client validation sets via secure aggregation, measuring calibration error (ECE) and task accuracy without any server-resident samples.

Have a different approach?

Describe how you would tackle this problem and we'll look for papers that already do it. Free; your text stays private.

Free · 3 checks per day

Why it might fail

If data-free server inversion leaks reconstructible private client samples via model weight inspection, the privacy advantage over holding public proxy data is invalidated. Alternatively, standard parameter-space regularizers may improve enough to close the performance gap without requiring alignment-based techniques at all.

Evidence

Each paper's own statement of the limitation, verbatim.

Nearest existing work

Related open problems

Federated Learning

Effect to explainOpen

Client-Scalable Federated Learning Under Severe Data Partitioning and Heterogeneity

Federated learning algorithms consistently exhibit severe accuracy collapses when the client population scales from small toy cohorts (5–10 clients) to realistic deployments (30–100+ clients). As the client count increases for a fixed dataset, each client receives fewer training samples while cross-client data heterogeneity and gradient conflict intensify, causing standard aggregation operators to produce destructive model updates. This scaling wall structurally limits federated systems from deploying in environments where data is naturally fragmented across large numbers of low-volume edge devices. Current literature masks this limitation by evaluating primarily on small, well-provisioned client counts where local sample sizes remain artificially high.

Strong candidate · 5/5 runs9 papers report this67% from 2025+
Generated automatically, not curated by hand. Automated prior-work checks catch about a third of existing work, so treat this problem as a lead to investigate.