Forward propagation 
So the forward propagation is ready and working. Unfortunately, due to memory constraints, I had to change the network structure a bit.

One 60x60 sample is processed by a 2x2 group of processors. Given that Parallella can have 4x4 or 8x8 grids, this allows 4 or 16 groups respectively work independently either on different samples, or on the same sample rotated (but that's for another post).

Now by layers:

Layer 0: each processor has 30x30 chunk of the image.

Layer 1: 16 5x5 filters are applied + 2x2 max-polling: output 14x14x16.

Layer 2: 32 5x5x16 filters + 2x2 max-polling: output 6x6x32.

Layer 3: 48 3x3x32 filters are applied + 2x2 max-polling: output 3x3x48 with overlap. The size of the full matrix at this stage is 5x5x48.

Layer 4: 64 3x3x48 convolutions. In the paper, this is unshared convolution, so that at each spot the weights are different. I am still not sure whether I will do it unshared, or replace by a regular convolution. TECHNICALLY the amount co computation on each core is unchanged, but the amount of data that has to be copied back and forth between ARM and Epiphany is ginormous... So regular convolution for now.
Also, since this layer a) doesn't have max-polling, b) is followed by fully-connected layer and c) the input is already so small in spatial dimensions, the partitioning of the output is a bit different now. Instead of each core applying the same filter to its chunk of data, it will now apply DIFFERENT filters to the FULL matrix. So, in the output each core will have a 3x3x16 chunk of a full 3x3x64 matrix.

Given that the output of every layer has to be preserved for the back-propagation, we have:

(30x30x1 + 14x14x16 + 6x6x32 + 3x3x48 + 3x3x16)*(4 bytes) = 22.5 kb

Out of 32 :(

Hope I have enough memory to do backpropagation without swapping out stuff :(

Comments

seo plugin 
Hello Web Admin, I noticed that your On-Page SEO is is missing a few factors, for one you do not use all three H tags in your post, also I notice that you are not using bold or italics properly in your SEO optimization. On-Page SEO means more now than ever since the new Google update: Panda. No longer are backlinks and simply pinging or sending out a RSS feed the key to getting Google PageRank or Alexa Rankings, You now NEED On-Page SEO. So what is good On-Page SEO?First your keyword must appear in the title.Then it must appear in the URL.You have to optimize your keyword and make sure that it has a nice keyword density of 3-5% in your article with relevant LSI (Latent Semantic Indexing). Then you should spread all H1,H2,H3 tags in your article.Your Keyword should appear in your first paragraph and in the last sentence of the page. You should have relevant usage of Bold and italics of your keyword.There should be one internal link to a page on your blog and you should have one image with an alt tag that has your keyword....wait there's even more Now what if i told you there was a simple Wordpress plugin that does all the On-Page SEO, and automatically for you? That's right AUTOMATICALLY, just watch this 4minute video for more information at. <a href="http://www.SEORankingLinks.com">Seo Plugin</a>

Add Comment
Fill out the form below to add your own comments.









Insert Special: