Spatio-temporal prediction plays an important role in many application areas especially in traffic domain. However, due to complicated spatio-temporal dependency and high non-linear dynamics in road networks, traffic prediction task is still challenging. Existing works either exhibit heavy training cost or fail to accurately capture the spatio-temporal patterns, also ignore the correlation between distant roads that share the similar patterns. In this paper, we propose a novel deep learning framework to overcome these issues: 3D Temporal Graph Convolutional Networks (3D-TGCN). Two novel components of our model are introduced. (1) Instead of constructing the road graph based on spatial information, we learn it by comparing the similarity between time series for each road, thus providing a spatial information free framework. (2) We propose an original 3D graph convolution model to model the spatio-temporal data more accurately. Empirical results show that 3D-TGCN could outperform state-of-the-art baselines.

In this paper, we propose a test, called Flagged-1-Bit (F1B) test, to study the intrinsic capability of recurrent neural networks in sequence learning. Four different recurrent network models are studied both analytically and experimentally using this test. Our results suggest that in general there exists a conflict between feature selection and feature memorization in sequence learning. Such a conflict can be resolved either using a gating mechanism as in LSTM, or by increasing the state dimension as in Vanilla RNN. Gated models resolve this conflict by adaptively adjusting their state-update equations, whereas Vanilla RNN resolves this conflict by assigning different dimensions different tasks. Insights into feature selection and memorization in recurrent networks are given.

In recent years, neural networks and deep learning have sparked tremendous progress in the field of natural language processing (NLP) and computer vision. While many of the face, object, landmark, logo, and text recognition and detection technologies are provided for Internet-connected devices, we believe that the ever-increasing computational power of mobile devices can enable the delivery of these technologies into the hands of users anytime, anywhere, regardless of Internet connection. However, computer vision for on-device and embedded applications faces many challenges -- models must run quickly with high accuracy in a resource-constrained environment, making use of limited computation, power, and space. TensorFlow offers various pre-trained models, such as drag-and-drop models, in order to identify approximately 1,000 default objects. When compared with other similar models, such as the Inception model datasets, MobileNet works better with latency, size, and accuracy.

Uber expanded Michelangelo "to serve any kind of Python model from any source to support other Machine Learning and Deep Learning frameworks like PyTorch and TensorFlow [instead of just using Spark for everything]." So why did Uber (and many other tech companies) build its own platform and framework-independent machine learning infrastructure? The posts How to Build and Deploy Scalable Machine Learning in Production with Apache Kafka and Using Apache Kafka to Drive Cutting-Edge Machine Learning describe the benefits of leveraging the Apache Kafka ecosystem as a central, scalable, and mission-critical nervous system. It allows real-time data ingestion, processing, model deployment, and monitoring in a reliable and scalable way. This post focuses on how the Kafka ecosystem can help solve the impedance mismatch between data scientists, data engineers, and production engineers. By leveraging it to build your own scalable machine learning infrastructure and also make your data scientists happy, you can solve the same problems for which Uber built its own ML platform, Michelangelo. Based on what I've seen in the field, an impedance mismatch between data scientists, data engineers, and production engineers is the main reason why companies struggle to bring analytic models into production to add business value.

We have seen a rise of AI technologies for image and video processing. Even though things tend to take a little while longer making it to the world of audio, here we have also seen impressive technological advances. In this article, I will summarize some of these advances, outline further potentials of AI in audio processing as well as describe some of the possible pitfalls and challenges we might encounter in pursuing this cause. The kicker for my interest in AI use cases for audio processing was the publication of Google Deepmind's "WaveNet" -- A deep learning model for generating audio recordings [1] which was released during the end of 2016. Using an adapted network architecture, a dilated convolutional neural network, Deepmind researchers succeeded in generating very convincing text-to-speech and some interesting music-like recordings trained from classical piano recordings.

Introduction: The number of glomeruli and glomerulosclerosis evaluated on kidney biopsy slides constitute as standard components of a renal pathology report. Prevailing methods for glomerular assessment remain manual, labor intensive and non-standardized. We developed a deep learning framework to accurately identify and segment glomeruli from digitized images of human kidney biopsies. Methods: Trichrome-stained images (n 275) from renal biopsies of 171 chronic kidney disease patients treated at the Boston Medical Center from 2009-12 were analyzed. A sliding window operation was defined to crop each original image to smaller images.

Posted by Jason A. Polzin, PhD GM Applications and Workflow, GE Healthcare Global Magnetic Resonance Imaging Here we describe our experience using TensorFlow to train a neural network to identify specific anatomy during a brain magnetic resonance imaging (MRI) exam to help improve speed and consistency. MRI (Figure 1.) is a 3D imaging technique that allows clinicians to visualize structures in the body non-invasively and without ionizing radiation. MRI is a widely used and powerful imaging modality due to its superior contrast between "soft" tissues, e.g. One of the key strengths of MRI is being able to image specific locations in the body at an orientation best suited for the purpose of the exam. This means that the operator must plan these scans carefully to yield the best possible images uniquely oriented for each patient to visualize the specific structures that may be of interest.

The Maathai Impact Award encourages and recognizes work by African innovators that show the impactful application of machine learning and artificial intelligence. The award will be presented at the annual Deep Learning Indaba in August 2019. This award reinforces the legacy of Wangari Maathai in acknowledging the capacity of individuals to be a positive force for change: by recognizing ideas and initiatives that demonstrate that each of us, no matter how small, can make a difference. In partnership with Black in AI, the winner will receive a fully-sponsored trip to attend NeurIPS 2019 and the Black in AI workshop, co-located with NeurIPS, in December 2019. The winner will also be invited to speak at the Deep Learning Indaba in Nairobi in August 2019 and receive a cash prize of KES 70,000.

The inadequacy of traditional forecasting model based on EEMD in practical work. For WTI, the first four IMFs decomposed by EEMD are suitable as inputs. Considering the actual demand of crude oil price forecasting, a novel model based on ensemble empirical mode decomposition (EEMD) and long short-term memory (LSTM) is proposed. In practical work, the model trained by historical data will be used in later data. Then the forecasting models based on EEMD need re-execute EEMD to update decomposition results of price series after getting new data.

The AI division of Google, known as DeepMind, claims its technology will boost the value of energy generated by onshore wind farms by 20%. Google purchases 2.6GW of renewable energy a year, close to 100% of its operational needs, and has agreements with 20 different wind and solar projects which could benefit from similar DeepMind analysis. Using a neural network (computer system inspired by the structure of biological networks) that was trained using weather forecasts and historical turbine data, DeepMind engineers configured their AI to predict power output of wind turbines 36 hours ahead of time. The benefits from using the AI in the management of the windfarm are threefold; better prediction of production, better prediction of power demands, and operational cost savings. DeepMind has been experimenting by applying its AI-led analysis to onshore turbines in the United States, on projects such as the Great Western Wind project in Oklahoma.