We present a scheme for implementing highly-connected, reconfigurable networks of integrate-and-fire neurons in VLSI. Neural activity is encoded by spikes, where the address of an active neuron is communicated through an asynchronous request and acknowledgement cycle. We employ probabilistic transmission of spikes to implement continuous-valued synaptic weights, and memory-based look-up tables ...

Topic models have proven versatile over the past decade, particularly as partial embeddings within more intricate models. These models present challenges that are analytic, computational and engineering in nature. Advances in probabilistic programming have the potential to circumvent a number of these issues, but researchers need a way to coarsely evaluate these frameworks. We identify three ax...

We propose a set of abstractions to modularize implementation of Bayesian inference algorithms. We provide a proof-of-concept implementation as a Haskell library and demonstrate on several examples how it simplifies implementation of Monte Carlo algorithms. Our technique is based on a method for modular construction of interpreters using monad transformers and is applicable generically to proba...

We present a new approach to learning probabilistic models for high dimensional data. This approach divides the data dimensions into low dimensional subspaces, and learns a separate mixture model for each subspace. The models combine in a principled manner to form a flexible modular network that produces a total density estimate. We derive and demonstrate an iterative learning algorithm that us...

Modular self-assembling on-orbit robots have the potential to reduce mission costs, increase reliability, and permit on-orbit repair and refueling. Modules with a variety of specialized capabilities would self-assemble from orbiting inventories. The assembled modules would then share resources such as power and sensors. As each free-flying module carries its own attitude control actuators, the ...

Swarm robotics studies large groups of robots that work together to accomplish common tasks. Much of the used source code is developed in an ad-hoc manner, meaning that the correctness of the controller is not always verifiable. In previous work, supervisory control theory (SCT) and associated design tools have been used to address this problem. Given a formal description of the swarm’s agents ...

The d-convex sets in a metric space are those subsets which include the metric interval between any two of its elements. Weak modularity is a certain interval property for triples of points. The d-convexity of a discrete weakly modular space X coincides with the geodesic convexity of the graph formed by the two-point intervals in X. The Helly number of such a space X turns out to be the same as...