AI Visualizations (Fusion Brain)
Relevant source files
The following files were used as context for generating this wiki page:
AI Visualizations within the ZINC-Fusion-V15 platform provide institutional-grade graphical representations of the multi-layer ensemble ML pipeline used for soybean oil (ZL) procurement forecasting. These components, primarily located in the frontend visualization suite, translate complex data structures—such as specialist signal correlations, neural connections, and market sentiment clusters—into interactive, organic interfaces for decision support.
The system utilizes specialized visual modules like the Fusion Brain, Neural Sphere, and Organic Topic Cloud to represent different layers of the intelligence stack. These visualizations allow users to monitor the "Big-11" specialist buckets, track regime shifts, and analyze the causal network of market drivers that influence the probabilistic multi-horizon forecasts.
Sources: README.md:12-25, frontend/src/app/page.tsx:14-25
Visualization Architecture
The visualization layer is built using modern web technologies including React, D3.js (d3-force), and Framer Motion. It is designed to handle real-time data from the Prisma Postgres backend, specifically from schemas like analytics, forecasts, and training.
Sources: AGENTS.md:12-25, frontend/src/components/viz/FusionBrain.tsx:4-7
Core Components
| Component | Purpose | Technical Implementation |
|---|---|---|
| Fusion Brain | Represents the causal network of market drivers and specialist correlations. | D3-force simulation with SVG synapses and React node components. |
| Neural Sphere | A high-level abstract representation of the core intelligence engine. | Three.js/React-based 3D particle sphere (referenced via props). |
| Organic Topic Cloud | Visualizes sentiment clusters and narrative volume from news sources. | D3-force simulation for organic bubble clustering and sentiment-based color mapping. |
Sources: frontend/src/components/viz/FusionBrain.tsx:10-25, frontend/src/components/viz/OrganicTopicCloud.tsx:13-30, frontend/src/app/page.tsx:32-35
Fusion Brain: Causal Network Mapping
The Fusion Brain component visualizes the "neural" connections between different market specialists. It maps the relationship between specialists such as Crush Economics, China Demand, Energy Complex, and Tariff Regimes.
Sources: frontend/src/components/viz/FusionBrain.tsx:28-35, Docs/zinc-fusion-v15-preview.html:1200-1250
Causal Force Simulation
The component uses a d3-force simulation to position nodes and links. Nodes represent specific market specialists or indicators, while links represent the correlation strength between them.
The simulation calculates distance based on the inverse of correlation value: distance = 200 * (1 - link.value).
Sources: frontend/src/components/viz/FusionBrain.tsx:50-65
Synapse Logic
Synapses (links) are rendered as SVG lines with active pulse animations reserved for "Strong" correlations (defined as value > 0.8).
Sources: frontend/src/components/viz/FusionBrain.tsx:90-110
Organic Topic Cloud: Sentiment Visualization
The Organic Topic Cloud translates qualitative sentiment data from the alt.* news tables into a dynamic cluster of "bubbles." This component visualizes the impact of news narratives on the market.
Sentiment Color Mapping
The component applies a specific color palette based on the sentiment value (-1 to 1):
- Bullish (> 0.12): Emerald/Green backgrounds with varying intensity.
- Bearish (< -0.12): Rose/Red backgrounds.
- Neutral (± 0.12): Slate/Grey backgrounds.
Sources: frontend/src/components/viz/OrganicTopicCloud.tsx:65-90
Clustering Mechanism
The visualization uses organic force layout to prevent overlapping while maintaining a centralized "cloud" structure.
Neural Sphere: Intelligence Abstraction
Used primarily on the landing page, the Neural Sphere provides a high-level visual cue for the "Institutional-Grade" nature of the platform. It accepts several configuration parameters to control its visual density and behavior.
Sources: frontend/src/app/page.tsx:32-40
Configuration Parameters
| Parameter | Type | Description |
|---|---|---|
size | Number | Overall diameter of the sphere. |
color | String | Base wireframe color. |
particleColor | String | Color of the individual neural data points. |
wireframeOpacity | Number | Transparency of the connecting geometry. |
rotationSpeed | Number | Velocity of the sphere's continuous rotation. |
Sources: frontend/src/app/page.tsx:34-39
Data Integration & Flow
The visualizations act as the final consumer in the ZINC-Fusion data pipeline.
The AI Intelligence layer provides the necessary reasoning and zlOutlook that influences the visual states (e.g., critical status in the Fusion Brain).
Sources: frontend/src/lib/ai-intelligence.ts:50-70, frontend/src/components/viz/FusionBrain.tsx:120-135
Summary
AI Visualizations in ZINC-Fusion-V15 serve as a sophisticated interface between the 19-model ensemble stack and the end-user. By utilizing D3-force simulations and 3D rendering, the platform provides a transparent view into the complex correlations of the "Big-11" specialists, allowing procurement teams to visually identify market regimes and sentiment-driven risks before they manifest in price action.
Sources: README.md:30-45, frontend/src/app/page.tsx:10-15