Architectural, governance-aware, and early-triage focused: bounded observation, provisional interpretation, short-horizon attentional escalation, and governed promotion.
Zenodo: zenodo.org/records/1960...
Repo: github.com/putmanmodel/...
Weak-Signal Interpretation for Early Cyber Anomaly Triage
A short bridge paper on how weak, partial cyber findings can influence attention before they harden into alarms, incident state, or ignored noise.
#Cybersecurity #ComputerSecurity #WeakSignals
Related repo:
github.com/putmanmodel/...
I add papers there regularly, often before public sharing.
#AIArchitecture #ReliabilityEngineering 3/3
This extends the same broader weak-signal / layered-interpretation architecture also used in recent bridge papers on drone detection, grid anomaly triage, and early cyber anomaly handling.
#AnomalyDetection #EarlyWarning 2/3
New report out:
Weak-Signal Interpretation for Industrial Fault Detection and Predictive Maintenance
A short bridge paper on handling weak equipment precursor signals before they harden into alarms, maintenance work, or ignored noise.
zenodo.org/records/1956...
#IndustrialAI #WeakSignals 1/3
This report is intentionally narrow: not a market-design paper, not a dispatch paper, and not a certified control claim.
It is an architectural bridge note showing that operator-facing grid anomaly triage is another serious home for weak-signal handling. (3/3)
Core pattern:
bounded finding → provisional candidate → attentional escalation → governed promotion
The point is not to ignore weak anomalies, and not to overread them either. The point is disciplined promotion control under uncertainty. (2/3)
#AnomalyDetection #SystemsArchitecture #PowerSystems
New report out:
Weak-Signal Architecture for Grid Anomaly Triage
A short bridge note on operator-facing early warning under uncertainty in grid and energy-system monitoring. (1/3)
zenodo.org/records/1956...
#EnergySystems #WeakSignals #EarlyWarning #ElectricalEngineering
The aim has been to keep the systems narrow, inspectable, and governed rather than overclaiming what they can know or do.
More of that work is on the way.
#AnomalyDetection #SystemsDesign #SignalProcessing
I’ve been pleasantly surprised by the response to the weak-signal drone detection paper.
It’s part of a broader set of bridge papers and projects using the same architectural approach in different domains.
#AI #DroneDetection
Also sharing a deterministic companion demo repo showing the layered behavior under synthetic scenarios:
github.com/putmanmodel/...
Broader architecture context:
github.com/putmanmodel/...
#AnomalyDetection #Simulation
3 of 3
Core architecture:
Sensor Partials → Reflex Salience → Deviation Evaluation → Candidate State → Governed Promotion
The goal is to let weak cues matter without letting them instantly become durable state or stronger consequence.
#SensorFusion #WeakSignal
2 of 3
Published: Weak-Signal Interpretation for Low-Altitude Drone Detection
A bridge paper arguing that low-altitude drone sensing is often better handled as a weak-signal interpretation problem than as a single-stage classification problem.
zenodo.org/records/1952...
#AI #DroneDetection
1 of 3
It includes 2 scenarios — “I’m fine.” and “Something smells weird.” — to show bounded observers, provisional interpretation, response shaping, and governance before consequence.
Repo: github.com/putmanmodel/...
Live: putmanmodel.github.io/notice-inter...
Paper: zenodo.org/records/1937...
I just published a small deterministic demo for a bounded weak-signal routing pattern:
Notice → Interpret → Shape → Govern
Weak cues should matter without instantly becoming truth, memory, or action.
#AI #Robotics #CognitiveScience #HCI #AgentSystems
Main idea: interactional cues can be treated more like weak physical signals.
Not magic. Not mind reading.
Just bounded representation, local comparison, provisional interpretation, and governed escalation before persistence or authority changes.
#CompanionSystems #CognitiveScience
A lot of systems fail the same way with weak human signals: they either ignore them, or overread them.
New preprint: From Signal Blends to Field Interpretation
A short argument for a narrower middle ground in AI interaction design.
#ArtificialIntelligence #HumanAIInteraction #AgentArchitectures
These are early, so my focus is on making them better before worrying too much about moat talk. If the work is strong enough, it has a better chance of finding the right people.
Paper: zenodo.org/records/1897...
S.U.N. Resolver is a small implementation-oriented companion to the paper: stable semantic object identity across perceptual routes, with ambiguity kept visible instead of collapsed into false certainty.
Released: S.U.N. Resolver
A small deterministic demo for stable semantic identity across changing perceptual routes.
Ambiguity stays visible. Behavior attaches to semantic identity, not raw perception.
github.com/putmanmodel/...
#ArtificialIntelligence #Robotics #CognitiveScience
Repo: github.com/putmanmodel/...
Companion paper: zenodo.org/records/1919...
Small, inspectable demo for context-dependent reconstruction, bridge-sensitive pruning, and rigidity effects.
Released: PUTMAN Interpretive Debugger
A narrow demo for contextual meaning reconstruction: same token, different context package, different result. Bridge relations matter. Rigidity can narrow what stays interpretable.
Live demo: putmanmodel.github.io/putman-inter...
#AI #CogSci #NarrativeDesign
Released: tvs-sdk v0.1
A lean SDK for TVS tags: typed grammar, versioned resolution, provenance, commit-class semantics, and deterministic replay.
Early-stage prototype. Looking for serious public feedback on use cases, failure modes, and adoption friction.
github.com/putmanmodel/...
#DevTools
Released: IP Transaction Governance Framework
doi.org/10.5281/zeno...
A governance-layer concept for IP deals before execution:
stages, gates, acknowledgements, pause logic, & auditability.
Includes a simple workflow simulator.
#IntellectualProperty #LegalTech #Licensing #DealFlow #RiskManagement
Released: CDE Lite v0.1
A narrow local prototype for conversational drift handling:
deterministic signals, scoped baselines, persistence, freeze behavior, evidence spans, and conservative routing.
Repo: github.com/putmanmodel/...
Feedback welcome.
#ConversationalAI #AIGovernance #NLP #OpenSource
Repo: github.com/putmanmodel/...
Small, controlled, and deterministic on purpose. If there’s real interest in this direction, I’ll keep building it out publicly.
Released: Guided Dialogue Trajectory
putmanmodel.github.io/guided-dialo...
A deterministic prototype for exploring how phrasing changes interpretation, scene state, and cue outputs. It also shows the direction I’m taking Scene Designer.
#GameDev #NarrativeDesign #DialogueSystems #IndieDev
Open source (CC BY-NC 4.0). Use it, extend it, or build on top of it.
This is one piece of a larger creative direction I’m building toward — tools that make dialogue, tone, and alternative conversational paths easier to explore directly.
Repo: github.com/putmanmodel/...
Released Dialogue Tone Workbench.
A deterministic, local-first tool for exploring how the same line changes meaning depending on tone.
Six controlled tone variants. Structured output. No external model dependency.
Try it: dialogue-tone-workbench.streamlit.app
#AI #OpenSource #GameDev #IndieDev