In this paper, we address the problem of reconstructing a time-domain signal (or a phase spectrogram) solely from a magnitude spectrogram. Since magnitude spectrograms do not contain phase information, we must restore or infer phase information to reconstruct a time-domain signal. One widely used approach for dealing with the signal reconstruction problem was proposed by Griffin and Lim. This method usually requires many iterations for the signal reconstruction process and depending on the inputs, it does not always produce high-quality audio signals. To overcome these shortcomings, we apply a learning-based approach to the signal reconstruction problem by modeling the signal reconstruction process using a deep neural network and training it using the idea of a generative adversarial network. Experimental evaluations revealed that our method was able to reconstruct signals faster with higher quality than the Griffin-Lim method.

How can we build agents that keep learning from experience, quickly and efficiently, after their initial training? Here we take inspiration from the main mechanism of learning in biological brains: synaptic plasticity, carefully tuned by evolution to produce efficient lifelong learning. We show that plasticity, just like connection weights, can be optimized by gradient descent in large (millions of parameters) recurrent networks with Hebbian plastic connections. First, recurrent plastic networks with more than two million parameters can be trained to memorize and reconstruct sets of novel, high-dimensional (1,000 pixels) natural images not seen during training. Crucially, traditional non-plastic recurrent networks fail to solve this task. Furthermore, trained plastic networks can also solve generic meta-learning tasks such as the Omniglot task, with competitive results and little parameter overhead. Finally, in reinforcement learning settings, plastic networks outperform a non-plastic equivalent in a maze exploration task. We conclude that differentiable plasticity may provide a powerful novel approach to the learning-to-learn problem.

There has been a lot of hype around data the past years. With the big data buzz cooling down, data now needs to be smart, apparently. Data scientists became the most sexy professionals alive, and got a martial arts assistant. Artificial intelligence has been hot for decades, the term seems to change meaning every now and then. Currently, machine and deep learning are the quickest rising data domains.

SEATTLE, April 05, 2018 (GLOBE NEWSWIRE) -- Applexus Technologies, a full-service business and technology solutions company based in the Seattle area, announced the launch of a new Artificial Intelligence (AI) practice to provide AI software and services to clients. The newly launched team is part of Applexus Product and Innovation. The new practice supports a wide variety of AI services and solutions, through custom and packaged solutions, using emerging technologies such as deep learning, machine learning and big data analysis. The Applexus Chief Technologist and AI practice leader is Dr. Thomas Koickal, one of the longest-serving global practitioners of machine learning based technologies. His career of more than 20 years has included work at the University of Edinburgh, UK and Vikram Sarabhai Space Centre, a research center of the Indian Space Research Organization.

This article is part of Demystifying AI, a series of posts that (try) to disambiguate the jargon and myths surrounding AI. In early 2016, Microsoft launched Tay, an AI chatbot that was supposed to mimic the behavior of a curious teenage girl and engage in smart discussions with Twitter users. The project would display the promises and potential of AI-powered conversational interfaces. However, in less than 24 hours, the innocent Tay became a racist, misogynist and a holocaust denying AI, debunking--once again--the myth of algorithmic neutrality. For years, we've thought that artificial intelligence doesn't suffer from the prejudices and biases of its human creators because it's driven by pure, hard, mathematical logic.

Using that you can create CNNs, RNNs, etc … on the browser and train these modules using the client's GPU processing power. Hence, a server GPU is not needed to train the NN. I created this simple demo with the code in Github. After this quick tutorial you should be able to understand the minimum requirements to create your first deep learning module in the browser. If you are familiar with deep learning platforms like TensorFlow you should be able to recognize that tensors are n dimensional arrays that are consumed by operators.

Motivation: Biomedical named entity recognition (BioNER) is the most fundamental task in biomedical text mining. State-of-the-art BioNER systems often require handcrafted features specifically designed for each type of biomedical entities. This feature generation process requires intensive labors from biomedical and linguistic experts, and makes it difficult to adapt these systems to new biomedical entity types. Although recent studies explored using neural network models for BioNER to free experts from manual feature generation, these models still require substantial human efforts to annotate massive training data. Results: We propose a multi-task learning framework for BioNER that is based on neural network models to save human efforts. We build a global model by collectively training multiple models that share parameters, each model capturing the characteristics of a different biomedical entity type. In experiments on five BioNER benchmark datasets covering four major biomedical entity types, our model outperforms state-of-the-art systems and other neural network models by a large margin, even when only limited training data are available. Further analysis shows that the large performance gains come from sharing character- and word-level information between different biomedical entities. The approach creates new opportunities for text-mining approaches to help biomedical scientists better exploit knowledge in biomedical literature.

In 2015, Google's DeepMind announced an advancement in creating an autonomous agent based on deep reinforcement learning (DRL) that could beat a professional player in a series of 49 Atari games. However, the current manifestation of DRL is still immature, and has significant drawbacks. One of DRL's imperfections is its lack of "exploration" during the training process, especially when working with high-dimensional problems. In this paper, we propose a mixed strategy approach that mimics behaviors of human when interacting with environment, and create a "thinking" agent that allows for more efficient exploration in the DRL training process. The simulation results based on the Breakout game show that our scheme achieves a higher probability of obtaining a maximum score than does the baseline DRL algorithm, i.e., the asynchronous advantage actor-critic method. The proposed scheme therefore can be applied effectively to solving a complicated task in a real-world application.

We conduct a large-scale study of language models for chord prediction. Specifically, we compare N-gram models to various flavours of recurrent neural networks on a comprehensive dataset comprising all publicly available datasets of annotated chords known to us. This large amount of data allows us to systematically explore hyper-parameter settings for the recurrent neural networks---a crucial step in achieving good results with this model class. Our results show not only a quantitative difference between the models, but also a qualitative one: in contrast to static N-gram models, certain RNN configurations adapt to the songs at test time. This finding constitutes a further step towards the development of chord recognition systems that are more aware of local musical context than what was previously possible.

We present an end-to-end trained memory system that quickly adapts to new data and generates samples like them. Inspired by Kanerva's sparse distributed memory, it has a robust distributed reading and writing mechanism. The memory is analytically tractable, which enables optimal on-line compression via a Bayesian update-rule. We formulate it as a hierarchical conditional generative model, where memory provides a rich data-dependent prior distribution. Consequently, the top-down memory and bottom-up perception are combined to produce the code representing an observation. Empirically, we demonstrate that the adaptive memory significantly improves generative models trained on both the Omniglot and CIFAR datasets. Compared with the Differentiable Neural Computer (DNC) and its variants, our memory model has greater capacity and is significantly easier to train.