Search

Many modern convolutional neural neworks (CNNs) rely on bottleneck block structures where the activation tensor is mapped between higher dimensions using an intermediate low dimension, and convolved with depthwise feature filters rather than multi-channel filters. Because most of the computation lies in computing the large dimensional tensors, however, such networks cannot be scaled without significant computation costs. In this paper, we show how \emph{fusing} the layers inside these blocks can dramatically reduce the multiplication count (by 6--20x) at the cost of extra additions. ReLU nonlinearities are predicted dynamically, and only the activations that survive ReLU contribute to directly compute the output of the block. We also propose FusioNet, a CNN architecture optimized for fusion, as well as ARCHON, a novel accelerator design with a dataflow optimized for fused networks. When FusioNet is executed on the proposed accelerator, it yields up to 5.8x faster inference compared to compact networks executed on a dense DNN accelerator, and 2.1x faster inference compared to the same networks when pruned and executed on a sparse DNN accelerator.

Coordination between ensembles of improvising electroacoustic musicians is a special case of the larger HCI problem of coordinating joint, real-time activity; one that involves some interesting additional and different challenges. This paper reports on research that has identified two specific real-time coordination problems for ensembles of electroacoustic musicians: "who makes what sound?" and "how is the sound being altered?" Real-time sound visualization is explored as a possible solution to assist musicians in overcoming some of these challenges. The main contribution of this paper is that, counterintuitively, for certain kinds of joint, real-time, coordination activities, temporal representations are important in helping to determine "who did what?"