My article entitled 'Deep neural networks show an equivalent and often superior performance to dermatologists in onychomycosis diagnosis: Automatic construction of onychomycosis datasets by region-based convolutional deep neural network' is published on January 19, 2018. We proposed the use of a region-based convolutional neural network (R-CNN) approach to created datasets of standardized nail images. It has been used to generate training datasets of approx. 50,000 images, which were then be able to fine-tune the ResNet-152 and VGG-19 models.
It was really fun getting to know tensorflow, but after I thought I knew enough I faced an even bigger challenge: What now? All of my hobby coding was always dedicated to solving some kind of problem I have, even if it's just automating 2 button presses into one. But with ML, I can't find any area of my life that could use it. Do you have any suggestions?
Would anyone have a good resource on the uses of word embeddings? I can see the awesomeness of word2vec to find words that are semantically 'close' but other than general information retrieval tasks I struggle to unerstand what word embeddings really do well. I would really like to see a list of applications and even products that use word embeddings; whether word2vec or otherwise.
Google has recently released a Jupyter Notebook platform called Google Colaboratory. You can run Python code in a browser, share results, and save your code for later. It currently does not support R code.
There're many environments where action spaces for the agent is variable or dependent on certain states the agent is in. For example in starcraft, dota 2, etc. What are some ways of dealing with this in the current RL framework? Use an RNN as action output for variable action output? Or delegate that to the environment where it just ignores or penalizes illegal actions?
Hello guys, I have to say I am quite new to the topic of machine learning. Still I am coming from a more mathematical/programming background. As a first project I would like to built a project that chooses for a text paragraph the appropriate meme. I was thinking of using a sentiment analysis library to get the mood of the paragraph and then use the giphy API to get the relevant gif. Do you have any further suggestions on the above presented implementation thoughts? Which libraries to you recommend for sentiment analysis? I would also highly appreciate any examples.