Goal In this ongoing summary we give a first introduction to the theano library, its basic functionality and usage in th field neural networks. Motivation The need for such a library is based on easily handling tensorial objects, like multi-array input data and weights in neural networks. Of particular interest are symbolic operations that can perform differentiation as need for the backpropagation algorithm in order to update the models according to the seen training data. A core functionality of libraries nowadays is to make use of GPUs in order to efficiently distribute the huge amount of computing operations. Ingredients symbolic operations, numpy, python Steps The theano library is an open source project lead by a machine learning group associated with […]
Goal The aim of this short summary is to provide an overview of the tasks an AI system is, should be and will be capable to execute in the near future. In particular we focus on classifying tasks with respect to NLP. Motivation When testing existing algorithms it is of great importance to have a clear interpretation of the results. This can be achieved by setting up specific learning environments in such a way that the complexity of the task is always under control. More generally, the state-of-the-art problems determine next development steps for Musio. Ingredients common sense, coreference, compound, conjunction, deduction, induction, argument relation, counting, negation, indefinite knowledge, time reasoning, positional and size reasoning, path finding, motivational reasoning. Steps […]
Today we would like to share a small video of us speaking to Musio about spaceships, hamburgers and life. In this small clip, we got to see what Musio thinks about them and have a little Q&A session and see answers from a robot’s point of view. All answers are reasoned out by the Artificial Intelligence engine powering Musio. It’s incredible, how much much Musio has learned over time. Check Musio below and tell us what you think!
Deep learning is a branch of Machine learning and is being actively explored in the recent years. Thanks to deep learning algorithms, it enables Musio to learn like us humans and let him do what he is capable of doing. Deep learning methods are better than the traditional machine learning and tech companies are always improving this area of technological advancements. The traditional machine learning techniques works in this way. If a program needed to be coded to recognise hand writing for instance, it’s quite a complex application to develop. The developer would need to devise rules on how characters are written, like ‘O’ is a rounded closed circle, but if someone writes it open on top, the application ‘thinks’ it’s a ‘U’ and here is […]
For more info : NN for Language Presenter : Juno This seminar introduces several popular technologies in deep learning or neural network for natural language processing such as distributed representation of word, neural network language model, recurrent neural network, and LSTM (Long short term memory). In particular, recent advances and a couple of application examples will also be discussed in the last part. – Reference: Using Neural Network for Modeling and Representing Natural Languages by Tomas Mikolov (Facebook) – Reference: Recurrent Nets and LSTM by Nando de Freitas (Oxford Univ.)