Presentation is loading. Please wait.

Presentation is loading. Please wait.

Towards Neuromorphic Complexity Analysis

Similar presentations


Presentation on theme: "Towards Neuromorphic Complexity Analysis"— Presentation transcript:

1 Towards Neuromorphic Complexity Analysis
Johan Kwisthout

2 Towards neuromorphic complexity analysis
Where do we fit? What sort of problems are efficiently solvable on a neuromorphic computer? Are these problems different / the same as the problems efficiently solvable on a Von Neumann architecture? Are we asking the right questions? What do we mean with “efficiently solvable” in this context? Time-efficient, space-efficient, or energy-efficient? Do our current models of computation (Turing machines) suffice to answer these questions?

3 Rationale Non-van Neumann architecture Energy as bounded resource Analog / spiking behavior Noise/randomness as resource

4 Rationale Non-van Neumann architecture Energy as bounded resource Analog / spiking behavior Noise/randomness as resource

5 Rationale Non-van Neumann architecture Energy as bounded resource Analog / spiking behavior Noise/randomness as resource

6 Rationale Non-van Neumann architecture Energy as bounded resource Analog / spiking behavior Noise/randomness as resource

7 Proposed computational framework
Key aspects are there: Co-located memory & computation Learning and adaptation Noise used as resource for sampling Spiking behavior Fruitful  firmly rooted in large body of theory

8 Open issues and research questions
Relationship with traditional models Probabilistic Turing machines Thresholds circuits with energy complexity Complexity classes, hardness criteria Existence of energy-hard problems? Structural complexity theory Reductions between classes Inclusions, proving hardness Notions of acceptance criteria Stability of distribution Time to convergence Energy limitations

9 Open for feedback/suggestions – work in progress!
Conclusion DoE 2016 workshop report, p. 29: “…likely that an entirely new computational theory paradigm will need to be defined in order to encompass the computational abilities of neuromorphic systems” Application: Describe what sort of problems can and cannot be solved tractably on neuromorphic hardware Describe fundamental limits of the brain’s (cognitive) capacity given the available resources Open for feedback/suggestions – work in progress!


Download ppt "Towards Neuromorphic Complexity Analysis"

Similar presentations


Ads by Google