MPC ModelsPrivate collaborative AI Training
Trained models

Cybersecurity threat detection: private training

1 row per party, 1 epoch, 13 trained features over a 28-feature threat schema. The participants do not get each other's source data; they get this trained scoring rule. The on-chain accounts below pin it to the recorded job — same schema, same training code, same revealed weights.

Stays private

Source rows and labels remain inside each participant's environment.

Gets verified

Party signatures, job state, schema hash, and final model hash.

Public log

Solana devnet stores the audit trail for this job.

Output

Logistic-regression rule · 28 weights + 1 bias.


What pushes the score up or down.

Each feature has one learned weight. Positive bars push the predicted score up; negative bars push it down. Inputs are normalized into [−1, 1] before MPC, so the magnitudes are directly comparable across columns.

LARGEST UPWARD PUSHES
TCP+1.6667e−3
ICMP+1.6667e−3
Failed Handshakes+1.6667e−3
Bytes Sent+1.6663e−3
Packets Sent+1.6663e−3
LARGEST DOWNWARD PUSHES
Src Port−1.6848e−3
UDP−1.6667e−3
Dst Port−3.8694e−4
FEATURE · RANGEDOWNWARDUPWARDWEIGHT
Duration Seconds
seconds
+1.6655e−3
Bytes Sent
bytes
+1.6663e−3
Bytes Received
bytes
+1.6660e−3
Packets Sent
packets
+1.6663e−3
Packets Received
packets
+1.6658e−3
TCP
binary
+1.6667e−3
UDP
binary
−1.6667e−3
ICMP
binary
+1.6667e−3
Src Port
port
−1.6848e−3
Dst Port
port
−3.8694e−4
Failed Handshakes
count
+1.6667e−3
Retransmissions
count
+1.6637e−3
Avg Packet Size
bytes
+1.5901e−3
Packet Size Variance
bytes_squared
0.0000e+0
Avg Interarrival Ms
milliseconds
0.0000e+0
Connections Last Hour
count
0.0000e+0
Src Country Code
numeric_country_code
0.0000e+0
Hour Of Day
hour
0.0000e+0
Day Of Week
day
0.0000e+0
Is Business Hours
binary
0.0000e+0
TLS version
version
0.0000e+0
Cert Valid
binary
0.0000e+0
Known CA
binary
0.0000e+0
Auth Failures
count
0.0000e+0
Session Resumption
binary
0.0000e+0
New Destination
binary
0.0000e+0
On Blocklist
binary
0.0000e+0
Volume Anomaly Score
score
0.0000e+0

Positive and negative mean direction in the learned score — not a security verdict. Bias term: -1.667e-3.


Accounts involved

Every public account and transaction below opens in Orb on devnet. The Arcium rows identify the computation definition and live computation account used by this MPC run.

The Solana devnet account that stores this receipt. Everything else hangs off of it.

The Anchor program that ties the MPC run to its commitments and finalizes the receipt on-chain.

Arcium wrapper programOpen program in Arcium Explorer

First-party Arcium Explorer page for the wrapper program that queued this computation.

The MPC training circuit that was executed for this job.

The on-chain list of parties allowed to join training runs.

Where each party's signed dataset commitment is stored.

The key used to verify the finalize signature for this job.

Arcium computation definition the MPC nodes executed.

The transaction that queued the MPC computation onto the cluster.

Arcium account that holds the in-flight MPC state for this job.

First-party Arcium Explorer page for this exact MPC computation run.

The transaction the cluster sent back with the finalized result. After this, the receipt is on-chain.

Run details
DomainCybersecurity threat detection
FinishedMay 27, 2026, 11:41 PM UTC
Circuittrain_logreg_leaf13
Training epochs1
Trained features13
Model features28
Rows per party1
Included parties bitset0b111
Circuit weight4,950,986,206 ACUs
Parties that joined
Aegis Industries Inc.US-CA
Nimbus Capital Pte LtdSG
Valor Health AGCH-ZH