AnomalyNet: files and card

README.md

@@ -0,0 +1,99 @@
+---
+license: apache-2.0
+library_name: pytorch
+pipeline_tag: tabular-classification
+tags:
+  - executorch
+  - zephyr
+  - tinyml
+  - anomaly-detection
+  - time-series
+  - industrial
+  - cortex-m
+  - int8
+datasets:
+  - eoinedge/heater-anomaly-windows
+metrics:
+  - accuracy
+  - precision
+  - recall
+model-index:
+  - name: AnomalyNet
+    results:
+      - task:
+          type: tabular-classification
+          name: Temperature-window anomaly detection
+        dataset:
+          type: eoinedge/heater-anomaly-windows
+          name: heater-anomaly-windows (test split)
+        metrics:
+          - type: accuracy
+            value: 0.9962
+          - type: precision
+            value: 1.0
+          - type: recall
+            value: 0.9925
+---
+
+# AnomalyNet
+
+An 842-parameter MLP that looks at 32 consecutive temperature samples (3.2 s at 10 Hz) and
+says whether the window is normal or contains an anomaly: a spike, a dropout, a runaway
+ramp or a burst of noise. It is the "second brain" in my dual-brain PLC demos: ExecuTorch
+runs it on the microcontroller next to a ladder-logic scan task or a Linux SoC, and only a
+shared variable (`p(anomaly)`) crosses between them.
+
+```
+fc1 32->24, ReLU
+fc2 24->2            logits [normal, anomaly]
+```
+
+Input: `(1, 32)` float, Celsius normalised as `(t - 50) / 20`. Output: 2 logits; softmax
+gives `p(anomaly)`.
+
+## Files
+
+| File | What |
+|---|---|
+| `anomaly_net.py` | model definition and the normalisation constants |
+| `anomaly_net.pt` | trained state dict, fp32 |
+| `executorch/fp32.pte` | ExecuTorch program, portable fp32 kernels (5.5 KB) |
+| `executorch/int8-cmsis-m0plus.pte`, `-m4.pte`, `-m33.pte`, `-m7.pte` | int8 via the ExecuTorch Cortex-M quantizer (3.6 KB) |
+| `tflite/fp32.tflite`, `tflite/int8.tflite` | the same weights through a Keras twin for TensorFlow Lite Micro |
+| `test_vectors.h` | one normal and one anomaly window with reference logits |
+
+## Results
+
+Held-out synthetic windows (dataset test split, 4,000 windows, 50 % anomalies):
+
+| Metric | Value |
+|---|---|
+| Accuracy | 99.62 % |
+| Precision | 100 % |
+| Recall | 99.25 % |
+| Parameters | 842 |
+| Inference, STM32U585 @ 160 MHz, fp32 | 1.6 ms |
+| Inference, nRF52833 @ 64 MHz, fp32 | 4.3 ms |
+| Inference, nRF9160 @ 64 MHz, fp32 | 4.0 ms |
+
+## Training
+
+Synthetic data only: the generator in `train/train_anomaly.py` of the
+[UNO Q repo](https://github.com/eoinjordan/executorch-zephyr-arduino-uno-q) simulates a
+heater loop around 50 C with slow drift and sensor noise, and injects four anomaly types.
+The same generator produced the
+[heater-anomaly-windows](https://huggingface.co/datasets/eoinedge/heater-anomaly-windows)
+dataset. Adam, cross-entropy, a few hundred epochs on CPU in under a minute.
+
+## Where it runs
+
+- [executorch-zephyr-arduino-uno-q](https://github.com/eoinjordan/executorch-zephyr-arduino-uno-q): STM32U585 + Linux side, 13x8 LED matrix chart
+- [executorch-zephyr-arduino-opta](https://github.com/eoinjordan/executorch-zephyr-arduino-opta): beside a ladder-logic scan task on a PLC
+- micro:bit v2, Thingy:91, LAUNCHXL-CC1352P1 self-tests, and the
+  [ExecuTorch vs TFLM benchmarks](https://github.com/eoinjordan/executorch-vs-tflm-benchmarks)
+
+Background: [AI for Industrial Control Systems: Bringing Edge AI to PLCs](https://www.edgeimpulse.com/blog/transform-industrial-control-systems-with-ai-anomaly-detection-made-easy/).
+
+## License
+
+Apache-2.0.

anomaly_net.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:556e735a12c213f1fdb79064655decdfa05ed9676969339c66e81bf481411bbf
+size 5829

anomaly_net.py

@@ -0,0 +1,34 @@
+# Copyright (c) 2026 Eoin Jordan
+# SPDX-License-Identifier: Apache-2.0
+"""AnomalyNet: a tiny MLP that classifies a window of temperature samples.
+
+Input: 32 consecutive readings (3.2 s at 10 Hz), normalised as (t - 50) / 20.
+Output: 2 logits, [normal, anomaly]. About 850 parameters, so the fp32
+portable ExecuTorch kernels run it in well under a millisecond on a
+Cortex-M7 and the whole program is a few KB of flash.
+"""
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+WINDOW = 32
+SAMPLE_HZ = 10
+MEAN = 50.0   # degrees C
+SCALE = 20.0  # degrees C
+CLASSES = ("normal", "anomaly")
+INPUT_SHAPE = (1, WINDOW)
+
+
+class AnomalyNet(nn.Module):
+    def __init__(self) -> None:
+        super().__init__()
+        self.fc1 = nn.Linear(WINDOW, 24)
+        self.fc2 = nn.Linear(24, len(CLASSES))
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        return self.fc2(F.relu(self.fc1(x)))
+
+
+def normalise(celsius: torch.Tensor) -> torch.Tensor:
+    return (celsius - MEAN) / SCALE

test_vectors.h

@@ -0,0 +1,12 @@
+/* Generated by models/export.py; do not edit. */
+#pragma once
+#define TEST_INPUT_NUMEL 32
+#define TEST_OUTPUT_NUMEL 2
+#define TEST_OUTPUT_ARGMAX 1
+static const float kTestInput[TEST_INPUT_NUMEL] = {
+	-0.0812852f, -0.1146014f, -0.1071035f, -0.0881012f, -0.1321746f, -0.0936299f, -0.1091498f, -0.0997824f,
+	-0.1035501f, -0.1003723f, -0.0939772f, -0.1167474f, -0.0850353f, -0.0964328f, -0.0956384f, -0.0917874f,
+	-0.0984082f, -0.0985181f, -0.1109982f, -0.8844818f, -0.8660494f, -0.8765051f, -0.8872373f, -0.8629481f,
+	-0.8762767f, -0.8841140f, -0.8852250f, -0.8664614f, -0.8660097f, -0.8524494f, -0.8733858f, -0.8584345f,
+};
+static const float kTestOutput[TEST_OUTPUT_NUMEL] = {-15.6083870f, 16.0947399f};

tflite/fp32.tflite

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e0e52c8e2e1907d8ef4abad5ca0a76f56727115a5bf2bcf299537a74989b335
+size 4832

tflite/int8.tflite

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6ed444b9ee05d928b2117f042e3bc7adcecde0f6e873d44382a6f703daea19a8
+size 3216