Share
Embed
Transcript
26 slides
-
Just a Neural Just a Neural Network! Network! "Deep learning"
refers to Deep Neural Networks A Deep Neural Network is simply a Neural Network with multiple hidden layers Neural Networks have been around since the 1970s -
Large Large Networks are Networks are hard to train hard
to train Vanishing gradients make backpropagation harder Overfitting becomes a serious issue So we settled (for the time being) with simpler, more useful variations of Neural Networks -
Then, Then, suddenly ... suddenly ... We realized we can
stack these simpler Neural Networks, making them easier to train We derived more efficient parameter estimation and model regularization methods Also, Moore's law kicked in and GPU computation became viable -
Speech Recognition Speech Recognition Baidu (with Andrew Ng as their
chief) has built a state- of-the-art speech recognition system with Deep Learning Their dataset: 7000 hours of conversation couple with background noise synthesis for a total of 100,000 hours They processed this through a massive GPU cluster -
Cross Domain Cross Domain Representations Representations What if you wanted
to take an image and generate a description of it? The beauty of representation learning is it's ability to be distributed across tasks This is the real power of Neural Networks -
Deep Learning NLP Deep Learning NLP Distributed word representations Dependency
Parsing Sentiment Analysis And many others ... -
Standard Standard Bag of Words A one-hot encoding 20k to
50k dimensions Can be improved by factoring in document frequency Word embedding Word embedding Neural Word embeddings Uses a vector space that attempts to predict a word given a context window 200-400 dimensions motel [0.06, -0.01, 0.13, 0.07, -0.06, -0.04, 0, -0.04] hotel [0.07, -0.03, 0.07, 0.06, -0.06, -0.03, 0.01, -0.05] Word Representations Word Representations Word embeddings make semantic similarity and synonyms possible -
Dependency Parsing Dependency Parsing Converting sentences to a dependency based
grammar Simplifying this to the verbs and it's agents is called Semantic Role Labeling -
Sentiment Sentiment Analysis Analysis Recursive Neural Networks Can model tree
structures very well This makes it great for other NLP tasks too (such as parsing) -
Get to the Get to the applications part applications part
already! already! -
Tools Tools Python Theano/PyLearn2 Gensim (for word2vec) nolearn (uses scikit-learn
Java/Clojure/Scala DeepLearning4j neuralnetworks by Ivan Vasilev APIs Alchemy API Meta Mind -
Problem: Funding Sentence Problem: Funding Sentence Classifier Classifier Build a
binary classifier that is able to take any sentence from a news article and tell if it's about funding or not. eg. "Mattermark is today announcing that it has raised a round of $6.5 million" -
Word Vectors Word Vectors Used Gensim's Word2Vec implementation to train
unsupervised word vectors on the UMBC Webbase Corpus (~100M documents, ~48GB of text) Then, iterated 20 times on text in news articles in the tech news domain (~1M documents, ~300MB of text) -
Sentence Vectors Sentence Vectors How can you compose word vectors
to make sentence vectors? Use paragraph vector model proposed by Quoc Le Feed into an RNN constructed by a dependency tree of the sentence Use some heuristic function to combine the string of word vectors -
What did we try? What did we try? TF-IDF +
Naive Bayes Word2Vec + Composition Methods Word2Vec + TF-IDF + Composition Methods Word2Vec + TF-IDF + Semantic Role Labeling (SRL) + Composition Methods -
Composition Methods Composition Methods Where wi represents the i'th word
vector, wv the word vector for the verb, and a0 and a1 are agents -
What worked? What worked? Word2Vec + TFIDF + SRL +
Circular Convolution /Additive The first method with simple TFIDF/Naive Bayes performed extremely poorly because of it's large dimensionality Combining TFIDF with Word2Vec provided a small, but noticeable improvement Adding SRL and a more sophisticated composition method increased performance by almost 5% -
What else could we try? What else could we try?
Can we apply this method to generate general purpose document vectors? We are currently using LDA (a topic analysis method) or simple TFIDF to create document vectors How will this method compare to the already proposed paragraph vector method by Quoc Le? Can we associate these document vectors with much smaller query strings? eg. Search for artificial intelligence against our companies and get better results than keyword search -
Who's doing ML at Who's doing ML at Mattermark? Mattermark?
mattermark We need more people! Refer anyone that you know that does Data Science/ML