2 ms·
> A large neural network can process sequences such as sentences while consuming four to sixteen times less energy while running on neuromorphic hardware than n
by mdp2021 4y ago
> A large neural network can process sequences such as sentences while consuming four to sixteen times less energy while running on neuromorphic hardware than non-neuromorphic hardware
> The group focused on algorithms that work with temporal processes ["stories comprehension"] ... The hardware tested consisted of 32 Loihi chips
> The group reproduced a presumed memory mechanism of the brain [... - ]the human brain can store information for a short period of time even without neural activity, namely in so-called "internal variables" of neurons. Simulations suggest that a fatigue mechanism of a subset of neurons is essential for this short-term memory. Direct proof is lacking because these internal variables cannot yet be measured, but it does mean that the network only needs to test which neurons are currently fatigued to reconstruct what information it has previously processed. In other words, previous information is stored in the non-activity of neurons, and non-activity consumes the least energy
Original research article: Arjun Rao et al, A Long Short-Term Memory for AI Applications in Spike-based Neuromorphic Hardware, https://dx.doi.org/10.1038/s42256-022-00480-w https://dx.doi.org/10.1038/s42256-022-00480-w