XNeuromorphic
ABYSON JOSEPH / INDEPENDENT PROJECT SPACE
I’m shaping research-led project concepts around spiking neural networks and event-driven AI: systems designed to respond to change with greater energy efficiency. Driven by an unwavering passion to challenge convention and explore what lies beyond today’s technology, I believe neuromorphic computing could transform how intelligent systems sense, respond, and adapt. I’m shaping applications for the Lava framework, targeting Intel Loihi—from anomaly detection to real-time perception and constrained optimization.
Independent exploration of brain-inspired computing, guided by a long-term vision for responsive, energy-efficient intelligence.
01 / UNDERSTANDING NEUROMORPHIC COMPUTING
Neuromorphic computing takes inspiration from how biological neurons communicate. In systems such as Intel Loihi, artificial neurons exchange brief signals called spikes, responding to events as they happen.
CONVENTIONAL AI
In conventional AI systems, processors and memory are typically separate, so data must move repeatedly between where it is stored and where calculations happen. This movement consumes energy and can become a performance bottleneck, particularly for demanding AI workloads.
NEUROMORPHIC COMPUTING
Neuromorphic architectures such as Intel Loihi bring memory and computation closer together. They also use event-driven processing, where artificial neurons exchange signals as events occur. This reduces unnecessary activity and data movement, creating opportunities for lower power consumption and faster responses.
As AI moves into robots, industrial equipment, and everyday devices, intelligence must work within tighter limits on energy, connectivity, and response time. Neuromorphic computing offers a promising way to address some of these constraints.
The opportunity is especially compelling for always-on sensing, real-time perception, and selected optimization tasks. Benefits depend on the application, hardware, and implementation; neuromorphic computing complements conventional AI and does not solve every AI limitation.
Intel’s research overview02 / PROJECT DIRECTIONS
From detecting subtle changes to making better decisions. These concepts define the applications I want to explore through neuromorphic computing.
SENSING & MONITORING
Identifying unusual patterns in streaming sensor data before they become difficult to ignore.
Equipment monitoring through changes in vibration, temperature, or electrical signals.
Use a recorded sensor dataset to compare a spiking neural network with a conventional anomaly-detection baseline.
Detection accuracy, false alarms, and response time. Hardware energy measurements would require a later Loihi evaluation.
Discuss this projectEVENT-BASED PERCEPTION
Following movement through changes in a scene, using the timing of visual events.
Responsive visual perception for robotics and systems that need to react to movement.
Build a single-object tracking pipeline using recorded event-camera data, with a view to a later Loihi implementation.
Tracking accuracy, response latency, and robustness across different movement patterns.
Discuss this projectCONSTRAINED OPTIMIZATION
Exploring how brain-inspired computation could assign work under real operational constraints.
Assigning tasks to limited resources while considering capacity, priorities, and deadlines.
Formulate a small scheduling problem using Lava’s optimization library and compare it with a conventional solver.
Solution quality, constraint satisfaction, and execution time as the problem size grows.
Discuss this projectCurrent stage: project concepts. Intel Loihi is the intended hardware target; no completed implementation or hardware results are claimed.
03 / THE PERSPECTIVE
I’m Abyson Joseph. Through XNeuromorphic, I’m defining independent neuromorphic project concepts around the Lava software framework, with Intel Loihi as the intended hardware target.
My focus is on three applications: detecting anomalies in sensor data, tracking movement through event-based vision, and scheduling tasks under operational constraints. Drawing on my software engineering and solution architecture experience, I’m shaping each concept into a practical prototype scope with clear integration requirements and evaluation criteria.
The next step is to implement and evaluate these concepts in Lava, then assess their suitability for Loihi hardware. I welcome discussions with researchers and engineers interested in the applications, implementation approach, or potential collaboration.
04 / TECHNICAL DIRECTION
An open-source framework for neuro-inspired applications. The proposed starting point for software experiments and optimization models.
Explore LavaNeuromorphic research hardware for event-driven computation. Hardware evaluation is a future step, subject to access and compatibility.
About Intel Loihi05 / OPEN CONVERSATION
A dataset. A use case. A different perspective.
I welcome conversations with researchers, engineers, and people interested in practical neuromorphic applications.
Ideas become more useful
when they meet another perspective.