The 150-sample Iris dataset used by the K-means article.

Includes standardization, K-means++ initialization, restarts, and SSE selection.

SVG flowchart for the C program execution path.

A 2D scatter projection using petal length and petal width.

Contains dataset, source code, flowchart, and visualization.

The article includes a runnable scikit-learn classification script.

A public advanced calculus practice PDF for review or printing.

The C source for the handwritten digit softmax classifier.

Compressed handwritten digit training set with 42000 labeled samples.

Compressed handwritten digit test set with 28000 unlabeled samples.

The official submission format example for checking the final output columns.

The prediction file generated by the current C project.

The compact softmax demo model and sample set used by the browser playground.

A small handwritten digit preview grid extracted from the training set.

Contains the source file, compressed datasets, submission files, browser model, and preview grid.

Single-file C tiny CNN with CIFAR-10 loading, convolution, pooling, softmax, and backpropagation.

Model weights generated by one local small-sample run.

Sample test prediction output from the CIFAR-10 tiny CNN.

Companion explanation material for the CNN project.

A redacted mld_chaffing_v2.py control-flow skeleton with secrets, node topology, and target lists removed.

A redacted CSV template for CPU, memory, peak threads, pulse rate, latency, and error measurements.

A CSV template for TP, FN, FP, TN, accuracy, precision, recall, F1, ROC-AUC, entropy, and JS divergence.

Notes explaining why the public resources include only redacted code, test templates, and architecture context.

Setup, safety boundaries, and quick-run commands for the AI Security series.

Includes safe toy scripts, result CSVs, risk register, attack-defense matrix, and architecture diagram.

CSV risk register template for AI threat modeling and release review.

Maps attack surface, toy demo, metric, and defensive control into one CSV table.

Shows threat modeling, robustness, data integrity, model privacy, and RAG guardrails.

FGSM-style perturbation and accuracy-drop experiment for a local digits classifier.

Demonstrates poison rate, trigger behavior, and attack success rate on digits.

Outputs membership AUC, target accuracy, surrogate fidelity, and surrogate accuracy.

Uses a deterministic toy agent to demonstrate external-data demotion and tool-policy blocking.

Setup, quick run, backend behavior, and output schemas for the PyBaMM battery AI data pipeline.

Bundles design generation, aging sweeps, EIS sweeps, label building, validation checks, sample CSVs, and figures.

Stores sample id, model family, parameter set, protocol, temperature, SOC, cycle, split group, and label source.

Frequency-level impedance output with frequency, Z_re, Z_im, magnitude, phase, backend, and solver status.

Stores SOH, RUL proxy, LLI, LAM, plating, local resistance, and EIS features.

Records duplicate samples, duplicate spectrum points, missing labels, split leakage, and backend usage.

Shows design grid, aging solve, EIS solve, label build, quality gate, and AI split.

Connects frequency points, impedance features, operating metadata, and SOH/RUL/degradation labels.

Sample figure showing cycle snapshots, SOH, and local ECM resistance labels.

Stores group split, MAE, RMSE, R2, label source, and backend used for auditing model results.

Stores held-out true values, predictions, and absolute errors.

Records random-forest feature importance values for each target model.

Shows held-out SOH/RUL prediction scatter plots and SOH feature importance.

OG share card for the PyBaMM battery modeling, EIS, aging simulation, and AI data hub.

Setup commands, script entry points, generated outputs, and figure notes for the math series.

Bundles NumPy scripts, CSV outputs, formula diagrams, loss contours, convolution figures, and attention heatmaps.

Stores MSE analytic gradients, finite-difference gradients, and error norms.

Step-by-step coordinates and loss for gradient descent, momentum, and Adam on a 2D quadratic.

Scores, softmax weights, and context vectors for a three-token scaled dot-product attention example.

Includes matrix shapes, computation graphs, loss contours, convolution scans, and attention heatmaps.

Browser modules for gradient checking, optimizer paths, convolution output size, and attention heatmaps.

Setup, no-privilege safety boundary, ten Python experiments, and three C examples.

Bundles Python/C source, fixed scenarios, ten result CSVs, and protocol/proxy figures.

HIT/MISS state, expiry, and latency for four fixed lookups.

Longest-prefix route and 3600-byte payload segmentation results.

Per-round ACK, window, and deterministic retransmission events.

Message direction, timing, and teaching shared value in a fixed RTT model.

Phase timing for HTTP/2 and HTTP/3 in cold and warm cache models.

Phase timing for direct access, forward-proxy tunneling, and reverse-proxy cache paths.

Records CONNECT authority, tunnel establishment, and the encrypted HTTPS-request boundary.

Stores ATYP, destination bytes, request length, and modeled local DNS counts.

Compares backend selection and queue waiting for round robin and least queue.

Records MISS, HIT, 304 revalidation, object age, and response latency.

Adjust TTL, prefixes, loss, handshake RTT, and cache paths in the browser.

Local reference script for character segmentation, PNG layers, and SVG vectorization.

Coordinates local ComfyUI, Grounding DINO, SAM, and vtracer to build SVG rig assets with stable IDs.

Editable API workflow template for IP-Adapter identity locking and ControlNet OpenPose exploded layout.

Layout reference for separated head, torso, arms, and legs; export to PNG for ControlNet if needed.

Grounding DINO prompts for semantic matting: head, left arm, torso, tail, accessory, and more.

A 1200x630 SVG card for the DNS, TLS, HTTP/3, proxy tunnel, and caching topic hub.

A 1200x630 SVG card for sharing the Deep Learning / CNN topic hub.

A 1200x630 SVG card for the K-means, Iris, and ML workflow topic hub.

A 1200x630 SVG card for eight queens, backtracking, bitmasks, and the playground.

A 1200x630 SVG card for handwritten digits, C classifiers, and browser demos.

Four 45-60 second storyboard scripts ready for later Remotion production.

An 8-second Remotion animation showing how a 3x3 convolution kernel scans an input and builds a feature map.

A Remotion clip showing sample assignment, centroid updates, and SSE reduction.

A Remotion clip showing row-by-row search, conflict pruning, and backtracking.