SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2206.07902 · NeurIPS · 2022

On Privacy and Personalization in Cross-Silo Federated Learning

Virginia Smith, Ziyu Liu, Steven Wu, Shengyuan Zhiwei

arXiv · PDF · Open in the Atlas

Code that ran

5 of 15 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 5 have no contract verdict; 0 failures of our instrument; 10 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2206.07902.json · badge: /api/ran/2206.07902.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
kenziyuliu/private-cross-silo-fl canonical (classifier label)
classifier basis not recorded on the link
5 ran · 10 no recorded run
FunctionStatusWhere it lives
client_selection Ran
ran on a synthesized input
kenziyuliu/private-cross-silo-fl/utils.py
code served (permissive licence) · get_code("8e1f359e172ce751")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 1, dtype family int, shape (10,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Set seed for reproducibility
np.random.seed(0)

# Define arguments for client_selection function
ARGS = {
    "seed": 42,
    "total": 100,
    "num_selected": 10,
    "weights": np.array([0.01] * 100)  # uniform weights normalized to sum to 1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family int, shape (10,) — shape (10,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
hinge_loss_hk Ran
ran on a synthesized input
kenziyuliu/private-cross-silo-fl/jax_utils.py
code served (permissive licence) · get_code("3ff20cad26a0cbfa")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ArrayImpl, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import jax
import jax.numpy as jnp
from jax import random
import haiku as hk

# Create a simple JAX model using Haiku
def forward_fn(inputs):
    x = hk.Linear(3)(inputs)
    x = jax.nn.relu(x)
    x = hk.Linear(1)(x)
    return x

# Initialize the model
key = random.PRNGKey(0)
dummy_input = jnp.ones((2, 4))
model = hk.transform(forward_fn)
params = model.init(key, dummy_input)

# Create a simple wrapper class that has the apply method
class SimpleModel:
    def __init__(self, transformed_model):
        self.transformed_model = transformed_model
    
    def apply(self, params, inputs):
        return self.transformed_model.apply(params, random.PRNGKey(0), inputs)

model_wrapper = SimpleModel(model)

# Create batch data
batch_inputs = jnp.ones((2, 4))
batch_targets = jnp.array([1.0, -1.0])
batch = (batch_inputs, batch_targets)

# Define the ARGS dictionary
ARGS = {
    'model': model_wrapper,
    'params': params,
    'batch': batch,
    'reg': 0.1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
haiku, jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ArrayImpl, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
read_gleam_data Ran
ran on a synthesized input
kenziyuliu/private-cross-silo-fl/utils.py
code served (permissive licence) · get_code("08e7e8d3e346c69d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import os
import numpy as np
import scipy.io
import tempfile

# Create a temporary directory for the mock data
temp_dir = tempfile.mkdtemp()

# Create a mock gleam.mat file with the expected structure
mock_data = {
    'X': np.array([
        np.array([
            np.random.randn(10, 5),  # 10 samples, 5 features for client 0
            np.random.randn(8, 5),   # 8 samples, 5 features for client 0
        ], dtype=object),
        np.array([
            np.random.randn(12, 5),  # 12 samples, 5 features for client 1
            np.random.randn(9, 5),   # 9 samples, 5 features for client 1
        ], dtype=object),
    ], dtype=object),
    'Y': np.array([
        np.array([
            np.random.randn(10, 1),  # 10 labels for client 0
            np.random.randn(8, 1),   # 8 labels for client 0
        ], dtype=object),
        np.array([
            np.random.randn(12, 1),  # 12 labels for client 1
            np.random.randn(9, 1),   # 9 labels for client 1
        ], dtype=object),
    ], dtype=object),
}

scipy.io.savemat(os.path.join(temp_dir, 'gleam.mat'), mock_data)

ARGS = {
    'data_dir': temp_dir,
    'seed': 42,
    'bias': False,
    'density': 1.0,
    'standardize': True,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 2, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
read_image Ran
ran on a synthesized input
kenziyuliu/private-cross-silo-fl/data_utils/adni_dataset.py
code served (permissive licence) · get_code("50f93eef977f4e82")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 3, dtype family float, shape (32, 32, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
from PIL import Image
import tempfile
import os

# Create a temporary image file for testing
temp_dir = tempfile.mkdtemp()
temp_image_path = os.path.join(temp_dir, "test_image.png")

# Create a simple 32x32 RGB image and save it
img_array = np.random.randint(0, 256, (32, 32, 3), dtype=np.uint8)
img = Image.fromarray(img_array, mode='RGB')
img.save(temp_image_path)

ARGS = {
    "path": temp_image_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family float, shape (32, 32, 3) — shape (32, 32, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
trunc_neg_binom_expectation Ran
ran on a synthesized input
kenziyuliu/private-cross-silo-fl/plot_hparam_cost.py
code served (permissive licence) · get_code("659cdbeeafbc296c")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

ARGS = {
    "eta": 0.5,
    "gamma": 0.7,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
adni_model_fn Not yet run
unverified (harvested, no recorded run)
kenziyuliu/private-cross-silo-fl/model_utils.py
code served (permissive licence) · get_code("d0ad384613d1ddd7")
How it ran

no invocation recorded.

bce_loss_hk Not yet run
unverified (harvested, no recorded run)
kenziyuliu/private-cross-silo-fl/jax_utils.py
code served (permissive licence) · get_code("8d57c71462cecacc")
How it ran

no invocation recorded.

linear_model_fn Not yet run
unverified (harvested, no recorded run)
kenziyuliu/private-cross-silo-fl/model_utils.py
code served (permissive licence) · get_code("1e50f08e3b522640")
How it ran

no invocation recorded.

mnist Not yet run
unverified (harvested, no recorded run)
kenziyuliu/private-cross-silo-fl/data_utils/rotated_mnist.py
code served (permissive licence) · get_code("ba51367756a7ee91")
How it ran

no invocation recorded.

mnist_model_fn Not yet run
unverified (harvested, no recorded run)
kenziyuliu/private-cross-silo-fl/model_utils.py
code served (permissive licence) · get_code("96103f2020a4840b")
How it ran

no invocation recorded.

read_data Not yet run
unverified (harvested, no recorded run)
kenziyuliu/private-cross-silo-fl/data_utils/adni_dataset.py
code served (permissive licence) · get_code("c9a99bc81bc03fae")
How it ran

no invocation recorded.

read_vehicle_data Not yet run
unverified (harvested, no recorded run)
kenziyuliu/private-cross-silo-fl/utils.py
code served (permissive licence) · get_code("fce14827432eb2fc")
How it ran

no invocation recorded.

rotated_mnist Not yet run
unverified (harvested, no recorded run)
kenziyuliu/private-cross-silo-fl/data_utils/rotated_mnist.py
code served (permissive licence) · get_code("50a1d1d5aed7afad")
How it ran

no invocation recorded.

rotated_patched_mnist Not yet run
unverified (harvested, no recorded run)
kenziyuliu/private-cross-silo-fl/data_utils/rotated_mnist.py
code served (permissive licence) · get_code("48ee242f558e091a")
How it ran

no invocation recorded.

sce_loss_hk Not yet run
unverified (harvested, no recorded run)
kenziyuliu/private-cross-silo-fl/jax_utils.py
code served (permissive licence) · get_code("5dca586694a038ce")
How it ran

no invocation recorded.

Also present in other papers' code

Of 15 samples lifted from this paper's linked repositories, 15 carry a function identity and 0 do not. Of the 15 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 15 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

While the application of differential privacy (DP) has been well-studied in crossdevice federated learning (FL), there is a lack of work considering DP and its implications for cross-silo FL, a setting characterized by a limited number of clients each containing many data subjects. In cross-silo FL, usual notions of client-level DP are less suitable as real-world privacy regulations typically concern the in-silo data subjects rather than the silos themselves. In this work, we instead consider an alternative notion of silo-specific sample-level DP, where silos set their own privacy targets for their local examples. Under this setting, we reconsider the roles of personalization in federated learning. In particular, we show that meanregularized multi-task learning (MR-MTL), a simple personalization framework, is a strong baseline for cross-silo FL: under stronger privacy requirements, silos are incentivized to federate more with each other to mitigate DP noise, resulting in consistent improvements relative to standard baseline methods. We provide an empirical study of competing methods as well as a theoretical characterization of MR-MTL for mean estimation, highlighting the interplay between privacy and cross-silo data heterogeneity. Our work serves to establish baselines for private cross-silo FL as well as identify key directions of future work in this area.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2206.07902")
get_code_for_paper("2206.07902")
have("2206.07902")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2206.07902.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2206.07902.svg)](https://syntology.ai/paper/2206.07902)

Connect an agent — have() is free.