SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.16738 · 2026

Self-Evolving Multi-Agent Network for Industrial IoT Predictive Maintenance

Truong-Son Hy, Rebin Saleh, Khanh Dinh, Balázs Villányi

arXiv · PDF · Open in the Atlas

Code that ran

3 of 4 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (2 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 2 have no contract verdict; 1 failure of our instrument; 1 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/2602.16738.json · badge: /api/ran/2602.16738.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
HySonLab/AgentIoT community
Papers with Code archive
3 ran · 1 no recorded run
FunctionStatusWhere it lives
FogPolicy Ran
ran on a synthesized input · constructed an object; no result was computed
HySonLab/AgentIoT/hama/agents/evolution.py
code served (permissive licence) · get_code("5206743d1028db04")
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 an instance of its own class, FogPolicy: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type 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 dataclasses import dataclass, field

@dataclass
class FogPolicy:
    """The locally adaptable policy parameters of one fog node."""
    w1: float = 0.5
    contamination: float = 0.3
    tau: float = 0.5

    def as_vector(self) -> np.ndarray:
        return np.array([self.w1, self.contamination, self.tau], dtype=float)

    @classmethod
    def from_vector(cls, v) -> 'FogPolicy':
        return cls(w1=float(v[0]), contamination=float(v[1]), tau=float(v[2]))


ARGS = {
    'w1': 0.75,
    'contamination': 0.25,
    'tau': 0.6,
}
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 FogPolicy — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a FogPolicy output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
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
PolicyEvolutionEnv Ran
ran on a synthesized input · constructed an object; no result was computed
HySonLab/AgentIoT/hama/agents/evolution.py
code served (permissive licence) · get_code("a7eaff0e02b65ca4")
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 an instance of its own class, PolicyEvolutionEnv: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type 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
import gymnasium as gym
from gymnasium import spaces

class MockSystem:
    """Mock system that simulates the fog computing system."""
    
    def __init__(self):
        self._global_policy = FogPolicy(w1=0.5, contamination=0.3, tau=0.5)
    
    def scores(self, X):
        """Return anomaly scores for input samples."""
        # Return scores between 0 and 1
        return np.random.default_rng(42).uniform(0, 1, len(X))
    
    def global_policy(self):
        """Return current global policy."""
        return self._global_policy
    
    def set_global_policy(self, policy, retrain_contamination=False):
        """Set the global policy."""
        self._global_policy = policy


class FogPolicy:
    """The locally adaptable policy parameters of one fog node."""
    w1: float = 0.5
    contamination: float = 0.3
    tau: float = 0.5

    def __init__(self, w1=0.5, contamination=0.3, tau=0.5):
        self.w1 = w1
        self.contamination = contamination
        self.tau = tau

    def as_vector(self):
        return np.array([self.w1, self.contamination, self.tau], dtype=float)

    @classmethod
    def from_vector(cls, v):
        return cls(w1=float(v[0]), contamination=float(v[1]), tau=float(v[2]))


# Create small validation data
np.random.seed(42)
X_val = np.random.randn(64, 4).astype(np.float32)  # 64 samples, 4 features
y_val = np.random.randint(0, 2, 64).astype(np.int32)  # Binary labels

# Create mock system
system = MockSystem()

ARGS = {
    'system': system,
    'X_val': X_val,
    'y_val': y_val,
    'window': 32,
    'episode_len': 8,
    'seed': 42,
    'retrain_contamination': False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
gymnasium, numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type PolicyEvolutionEnv — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a PolicyEvolutionEnv output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
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
classification_metrics Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
HySonLab/AgentIoT/hama/agents/evolution.py
code served (permissive licence) · get_code("34a9706c22c472e6")
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 dict. The image it ran in was not recorded. If your copy returns the same type 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 random seed for reproducibility
np.random.seed(0)

# Create sample data for classification metrics
# y_true: binary labels (0 or 1)
y_true = np.array([0, 1, 1, 0, 1, 0, 1, 1, 0, 0])

# scores: prediction scores (continuous values between 0 and 1)
scores = np.array([0.1, 0.9, 0.8, 0.2, 0.85, 0.15, 0.95, 0.88, 0.3, 0.25])

# threshold: classification threshold
threshold = 0.5

ARGS = {
    'y_true': y_true,
    'scores': scores,
    'threshold': threshold
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
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
evolve_policy Not yet run
unverified (harvested, no recorded run)
HySonLab/AgentIoT/hama/agents/evolution.py
code served (permissive licence) · get_code("707adc99bdf365e2")
How it ran

no invocation recorded.

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 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 4 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

Industrial IoT predictive maintenance requires systems capable of real-time anomaly detection without sacrificing interpretability or demanding excessive computational resources. Traditional approaches rely on static, offline-trained models that cannot adapt to evolving operational conditions, while LLM-based monolithic systems demand prohibitive memory and latency, rendering them impractical for on-site edge deployment. We introduce SEMAS, a self-evolving hierarchical multi-agent system that distributes specialized agents across Edge, Fog, and Cloud computational tiers. Edge agents perform lightweight feature extraction and pre-filtering; Fog agents execute diversified ensemble detection with dynamic consensus voting; and Cloud agents continuously optimize system policies via Proximal Policy Optimization (PPO) while maintaining asynchronous, non-blocking inference. The framework incorporates LLM-based response generation for explainability and federated knowledge aggregation for adaptive policy distribution. This architecture enables resource-aware specialization without sacrificing real-time performance or model interpretability. Empirical evaluation on two industrial benchmarks (Boiler Emulator and Wind Turbine) demonstrates that SEMAS achieves superior anomaly detection performance with exceptional stability under adaptation, sustains prediction accuracy across evolving operational contexts, and delivers substantial latency improvements enabling genuine real-time deployment. Ablation studies confirm that PPO-driven policy evolution, consensus voting, and federated aggregation each contribute materially to system effectiveness. These findings indicate that resource-aware, self-evolving 1multiagent coordination is essential for production-ready industrial IoT predictive maintenance under strict latency and explainability constraints. Our implementation is publicly available at https://github.com/HySonLab/AgentIoT.

For agents

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

get_harvested_code_for_paper("2602.16738")
get_code_for_paper("2602.16738")
have("2602.16738")

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

curl https://syntology.ai/api/ran/2602.16738.json

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

[![Syntology run record](https://syntology.ai/api/ran/2602.16738.svg)](https://syntology.ai/paper/2602.16738)

Connect an agent — have() is free.