Title | ||
---|---|---|
Efficient Convolutional Neural Networks For Pixelwise Classification On Heterogeneous Hardware Systems |
Abstract | ||
---|---|---|
With recent advances in high-throughput Electron Microscopy (EM) imaging it is now possible to image an entire nervous system of organisms like Drosophila melanogaster. One of the bottlenecks to reconstruct a connectome from these large volumes (approximate to 100 TiB) is the pixel-wise prediction of membranes. The time it would typically take to process such a volume using a convolutional neural network (CNN) with a sliding window approach is in the order of years on a current GPU. With sliding windows, however, a lot of redundant computations are carried out. In this paper, we present an extension to the Caffe library to increase throughput by predicting many pixels at once. On a sliding window network successfully used for membrane classification, we show that our method achieves a speedup of up to 57x, maintaining identical prediction results. |
Year | DOI | Venue |
---|---|---|
2015 | 10.1109/ISBI.2016.7493487 | 2016 IEEE 13TH INTERNATIONAL SYMPOSIUM ON BIOMEDICAL IMAGING (ISBI) |
Keywords | Field | DocType |
convolutional neural networks, pixel wise classification, electron microscopy, loss functions | Histogram,Convolutional neural network,Computer science,Input/output,Artificial intelligence,Artificial neural network,Computer hardware,Sliding window protocol,Pattern recognition,Softmax function,Caffè,Pixel,Machine learning | Journal |
Volume | ISSN | Citations |
abs/1509.03371 | 1945-7928 | 0 |
PageRank | References | Authors |
0.34 | 0 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
fabian tschopp | 1 | 0 | 0.34 |
Julien N P Martel | 2 | 28 | 7.46 |
Srinivas C. Turaga | 3 | 127 | 23.75 |
Matthew Cook | 4 | 100 | 10.77 |
Jan Funke | 5 | 50 | 7.48 |