The input is a single text file and the model learns to predict the next character in the sequence. Hoping to see many public char-rnn micro-api s with different models spring up, so we can experiment together more easily. This code implements multi-layer Recurrent Neural Network (RNN, LSTM, and GRU) for training/sampling from character-level language models. The input is a single text file and the model learns to predict the next character in the sequence. The context of this code base is described in detail in my blog post.

Based on its core scientist team's top-level R&D strength, as well as technological innovation and breakthroughs, Squirrel AI Learning started holding four "man-machine competitions" in Zhengzhou, Chengdu and Dongying in October 2017 in a bid to identify any difference between its adaptive learning system and human teaching. Dr. Kalns demonstrated to the RE-WORK audience the results of the four competitions: surprisingly, machine teaching outperformed human teaching in all the four competitions. Taking the fourth competition, which unfolded in one hundred cities, as an example, students at the same intellectual level were divided into two groups and received human teaching and Squirrel AI Learning respectively. Every student in the machine teaching group learned 42 knowledge points on the average, while every student in the human teaching learned 28 knowledge points on the average; in terms of average scoring in the core part of the competition, the students in the AI teaching group had their scores increased by 5.4 on the average, while the students in the human teaching group just had their scores increased by 0.7 on the average, suggesting that machine teaching enabled students to take a firmer grasp of knowledge points than human teaching and improved the learning efficiency more significantly than human teaching. According to the results, Squirrel AI Learning is basically the same as or better than individualized human teaching.

It's critical to understand that AI is not one technology. It is a set of technologies and building blocks that include natural language processing, machine and deep learning, neural networks, virtual agents, autonomics and computer vision. How can IBM help you?

Computer-aided breast cancer diagnosis in mammography is limited by inadequate data and the similarity between benign and cancerous masses. To address this, we propose a signed graph regularized deep neural network with adversarial augmentation, named \textsc{DiagNet}. Firstly, we use adversarial learning to generate positive and negative mass-contained mammograms for each mass class. After that, a signed similarity graph is built upon the expanded data to further highlight the discrimination. Finally, a deep convolutional neural network is trained by jointly optimizing the signed graph regularization and classification loss. Experiments show that the \textsc{DiagNet} framework outperforms the state-of-the-art in breast mass diagnosis in mammography.

Next-generation wireless networks must support ultra-reliable, low-latency communication and intelligently manage a massive number of Internet of Things (IoT) devices in real-time, within a highly dynamic environment. This need for stringent communication quality-of-service (QoS) requirements as well as mobile edge and core intelligence can only be realized by integrating fundamental notions of artificial intelligence (AI) and machine learning across the wireless infrastructure and end-user devices. In this context, this paper provides a comprehensive tutorial that introduces the main concepts of machine learning, in general, and artificial neural networks (ANNs), in particular, and their potential applications in wireless communications. For this purpose, we present a comprehensive overview on a number of key types of neural networks that include feed-forward, recurrent, spiking, and deep neural networks. For each type of neural network, we present the basic architecture and training procedure, as well as the associated challenges and opportunities. Then, we provide an in-depth overview on the variety of wireless communication problems that can be addressed using ANNs, ranging from communication using unmanned aerial vehicles to virtual reality and edge caching.For each individual application, we present the main motivation for using ANNs along with the associated challenges while also providing a detailed example for a use case scenario and outlining future works that can be addressed using ANNs. In a nutshell, this article constitutes one of the first holistic tutorials on the development of machine learning techniques tailored to the needs of future wireless networks.

Time series forecasting is an important problem across many domains, including predictions of solar plant energy output, electricity consumption, and traffic jam situation. In this paper, we propose to tackle such forecasting problem with Transformer. Although impressed by its performance in our preliminary study, we found its two major weaknesses: (1) locality-agnostics: the point-wise dot-product self attention in canonical Transformer architecture is insensitive to local context, which can make the model prone to anomalies in time series; (2) memory bottleneck: space complexity of canonical Transformer grows quadratically with sequence length $L$, making modeling long time series infeasible. In order to solve these two issues, we first propose convolutional self attention by producing queries and keys with causal convolution so that local context can be better incorporated into attention mechanism. Then, we propose LogSparse Transformer with only $O(L(\log L)^{2})$ memory cost, improving the time series forecasting in finer granularity under constrained memory budget. Our experiments on both synthetic data and real-world datasets show that it compares favorably to the state-of-the-art.

Nowadays, users open multiple accounts on social media platforms and e-commerce sites, expressing their personal preferences on different domains. However, users' behaviors change across domains, depending on the content that users interact with, such as movies, music, clothing and retail products. In this paper, we propose an adaptive deep learning strategy for cross-domain recommendation, referred to as ADC. We design a neural architecture and formulate a cross-domain loss function, to compute the non-linearity in user preferences across domains and transfer the knowledge of users' multiple behaviors, accordingly. In addition, we introduce an efficient algorithm for cross-domain loss balancing which directly tunes gradient magnitudes and adapts the learning rates based on the domains' complexities/scales when training the model via backpropagation. In doing so, ADC controls and adjusts the contribution of each domain when optimizing the model parameters. Our experiments on six publicly available cross-domain recommendation tasks demonstrate the effectiveness of the proposed ADC model over other state-of-the-art methods. Furthermore, we study the effect of the proposed adaptive deep learning strategy and show that ADC can well balance the impact of the domains with different complexities.

Census data provide detailed information about population characteristics at a coarse resolution. Nevertheless, fine-grained, high-resolution mappings of population counts are increasingly needed to characterize population dynamics and to assess the consequences of climate shocks, natural disasters, investments in infrastructure, development policies, etc. Dissagregating these census is a complex machine learning, and multiple solutions have been proposed in past research. We propose in this paper to view the problem in the context of the aggregate learning paradigm, where the output value for all training points is not known, but where it is only known for aggregates of the points (i.e. in this context, for regions of pixels where a census is available). We demonstrate with a very simple and interpretable model that this method is on par, and even outperforms on some metrics, the state-of-the-art, despite its simplicity.

We deal with the \textit{selective classification} problem (supervised-learning problem with a rejection option), where we want to achieve the best performance at a certain level of coverage of the data. We transform the original $m$-class classification problem to $(m+1)$-class where the $(m+1)$-th class represents the model abstaining from making a prediction due to uncertainty. Inspired by portfolio theory, we propose a loss function for the selective classification problem based on the doubling rate of gambling. We show that minimizing this loss function has a natural interpretation as maximizing the return of a \textit{horse race}, where a player aims to balance between betting on an outcome (making a prediction) when confident and reserving one's winnings (abstaining) when not confident. This loss function allows us to train neural networks and characterize the uncertainty of prediction in an end-to-end fashion. In comparison with previous methods, our method requires almost no modification to the model inference algorithm or neural architecture. Experimentally, we show that our method can identify both uncertain and outlier data points, and achieves strong results on SVHN and CIFAR10 at various coverages of the data.

Artificial intelligence and machine learning networks could help personalize radiation therapy for lung cancer, according to a new study by the Cleveland Clinic. The research, published in The Lancet Digital Health, centers around an artificial neural network built with a large dataset of patients receiving lung radiotherapy. That network, which allows each clinical center to utilize their own CT datasets to customize the framework and tailor it to their specific patient population, was built using CT scans and the electronic health records of nearly a thousand lung cancer patients treated with high-dose radiation. The company's framework uses probability estimates to select an optimized dose that prevents treatments failures to a set level, for instance a five percent probability of failure. Pre-treatment scans were input into a deep-learning model, which analyzed the scans to create an image signature that predicts treatment outcomes.