Neural Network Companion
Final Year Project
Brunel University - Independant Researcher and Developer
Neural Network Companion Case Study
Neural Network Companion is a final-year research and development project exploring emotionally aware AI through facial-expression recognition, natural language interaction, and adaptive response design.
Problem
Most conversational systems respond only to text, which means they miss non-verbal emotional signals that shape real human communication. This project investigated how a digital companion could use facial cues alongside language to create more context-aware interactions.
Role
As the Independent Researcher and Developer, I researched the problem space, designed the prototype architecture, trained and integrated recognition components, and built the interaction flow for the companion experience.
Architecture
- Perception: Facial-expression recognition analyzes visual input to infer broad emotional states.
- Conversation: NLP components support responsive dialogue that can adapt tone and context based on detected cues.
- AI Tooling: The project explored custom neural-network work alongside external AI libraries and pre-trained model concepts.
- Extensibility: The system was structured as a proof of concept that could later support richer personalization and more advanced companion behaviors.
Stack
Python, neural networks, facial-expression recognition, natural language processing, machine-learning experimentation, and emotionally aware interaction design.
Media
The project page includes an embedded Vimeo prototype video demonstrating the companion concept and interaction direction.
Outcome / Impact
NNC demonstrates an applied research approach to multimodal AI, combining computer vision, NLP, and user-centered interaction design. It is presented as a proof of concept for emotionally responsive systems, not a clinical or therapeutic product.
Tags: artificial intelligence, ai, tensorflow, deepmind, ibm, software, programming, developer, engineer, Neural Network Companion, NNC, gsap, webgl, webdev, gamedev, threejs, 3d, vfx
Joshua Fields — full portfolio
