Federated Training Control Deck
Configure datasets, select custom features and labels in tabular format, calibrate edge device enclaves with equal or manual dataset splits, design neural hidden layers with unlimited epochs, and inspect live multi-node logs.
Dataset Selection
Choose a synthetic test dataset preset or enter / upload your own CSV data.
Healthcare EHR Diagnostics
Multivariate patient diagnostic records with clinical biomarkers for early pathology detection.
Genomic Multi-Omics Sequencing (Deep Cancer Profiling)
High-throughput 32-biomarker somatic mutation profiling and next-generation RNA-seq expression telemetry. Heavy multi-parameter convergence across distributed medical enclaves.
High-Frequency Order Book Microstructure (L2 Telemetry)
Sub-millisecond institutional order book depth, cross-exchange liquidity imbalances, and volatility spikes across 24 latency-sensitive signals.
Financial Transaction Anomaly
Credit card telemetry, transaction velocity, risk scores, and geospatial signals for fraud detection.
Industrial IoT Edge Telemetry
Multi-sensor industrial telemetry monitoring rotor temperature, acoustic vibrations, and RPM deviations.
MNIST Digit PCA Embeddings
16-dimensional PCA eigenprojections of handwritten digits for distributed privacy-preserving classification.
Features & Target Tabular Profiler
Inspect columns in tabular format showing unique values, hot/cold variance signal, memory size, and designate target label.
| Include | Column Name | Role | Data Type | Unique Values | Feature Encoding | Memory Size | Missing | Value Range / Sample | Action |
|---|---|---|---|---|---|---|---|---|---|
| diagnostic_risk | TARGET | integer | 2 unique | 3.8 KB | 0 (0%) | [0, 1] Binary | Active Target | ||
| age | FEATURE | numeric | 52 unique | 3.8 KB | 0 (0%) | 24 - 86 yrs | |||
| systolic_bp | FEATURE | numeric | 68 unique | 3.8 KB | 0 (0%) | 98 - 188 mmHg | |||
| diastolic_bp | FEATURE | numeric | 44 unique | 3.8 KB | 0 (0%) | 60 - 115 mmHg | |||
| fasting_glucose | FEATURE | numeric | 94 unique | 3.8 KB | 0 (0%) | 65 - 240 mg/dL | |||
| cholesterol_ldl | FEATURE | numeric | 112 unique | 3.8 KB | 0 (0%) | 80 - 245 mg/dL | |||
| troponin_i | FEATURE | numeric | 38 unique | 3.8 KB | 0 (0%) | 0.01 - 0.45 ng/mL | |||
| crp_biomarker | FEATURE | numeric | 49 unique | 3.8 KB | 0 (0%) | 0.2 - 14.8 mg/L | |||
| wbc_count | FEATURE | numeric | 58 unique | 3.8 KB | 0 (0%) | 4.1 - 16.5 k/uL | |||
| creatinine_serum | IGNORED | numeric | 31 unique | 3.8 KB | 0 (0%) | 0.6 - 2.8 mg/dL | |||
| bmi_index | IGNORED | numeric | 76 unique | 3.8 KB | 0 (0%) | 18.2 - 42.6 | |||
| spo2_deficit | IGNORED | numeric | 22 unique | 3.8 KB | 0 (0%) | 0 - 8 % | |||
| respiratory_rate | IGNORED | numeric | 18 unique | 3.8 KB | 0 (0%) | 12 - 28 bpm | |||
| smoker_status | IGNORED | categorical | 3 unique | 3.8 KB | 0 (0%) | [0: Never, 1: Former, 2: Active] | |||
| genetic_marker_a | IGNORED | boolean | 2 unique | 3.8 KB | 0 (0%) | [0, 1] | |||
| genetic_marker_b | IGNORED | boolean | 2 unique | 3.8 KB | 0 (0%) | [0, 1] |
Participating Edge Devices & Data Split (0 Active)
Choose which nodes participate and specify whether dataset samples are equally divided or manually allocated.
No Edge Devices Connected
By default, no mock nodes are present. Connect an actual physical laptop using the official worker script above, or spawn simulated test nodes on-demand to test locally.
Python Pipeline Studio & Distributed Execution
Design custom model architectures & local training steps. Edge nodes run the pipeline on their local private datasets with zero raw data sharing.
Each connected edge device runs the same pipeline logic locally on its own private hardware enclave. Devices can specify their own local dataset path (--data-path) and edit pipeline.py locally without affecting other nodes!
python fedzero_worker.py --server http://localhost:8000 --name "Hospital_Beta_Node" --data-path "C:/data/patient_records.csv" --pipeline "./pipeline.py"Neural Hidden Layers & Hyperparameters
Add or remove hidden layers, customize neurons per layer, enter any number of epochs, and set custom learning rate.
Matches the 8 active features selected from table.
Objective target: diagnostic_risk.
Enter any number of epochs and custom learning rate, then start the training round.
Multi-Node Distributed Logs Terminal
Simultaneous execution streams for all participating edge nodes (0 active devices).