preview

Deep Generative Model Of Artificial System

Decent Essays
Open Document
Deep generative models (i.e., generative models implemented as multi-layered neural networks) have recently shown striking successes in producing synthetic outputs that capture the form and structure of real visual scenes via the incorporation of attention-like mechanisms (Hong et al., 2015; Reed et al., 2016). For example, in one state-of-the-art generative model known as DRAW, attention allows the system to build up an image incrementally, attending to one portion of a “mental canvas” at a time (Gregor et al., 2015).
The Future
In AI, the pace of recent research has been remarkable. Artificial systems now match human performance in challenging object recognition tasks (Krizhevsky et al., 2012) and outperform expert humans in dynamic,…show more content…
For example, when paused at a choice point, ripples of neural activity in the rat hippocampus resemble those observed during subsequent navigation of the available trajectories (“preplay”), as if the animal were “imagining” each possible alternative (Johnson and Redish, 2007; Ólafsdóttir et al., 2015; Pfeiffer and Foster, 2013). Further, recent work has suggested a similar process during non-spatial planning in humans (Doll et al., 2015; Kurth-Nelson et al., 2016). We have discussed above the ways in which the introduction of mechanisms that replay and learn offline from past experiences can improve the performance of deep RL agents such as DQN.
Machine learning techniques have transformed the analysis of neuroimaging datasets—for example, in the multivariate analysis of fMRI and magnetoencephalographic (MEG) data (Cichy et al., 2014; Çukur et al., 2013; Kriegeskorte and Kievit, 2013)—with promise for expediting connectomic analysis (Glasser et al., 2016), among other techniques. Going further, we believe that building intelligent algorithms has the potential to offer new ideas about the underpinnings of intelligence in the brains of humans and other animals. In particular, psychologists and neuroscientists often have only quite vague notions of the
Get Access
Related
Get Access