Researchers from Facebook, the National University of Singapore, and the Qihoo 360 AI Institute have jointly proposed OctConv (Octave Convolution), a promising new alternative to traditional convolution operations. Akin to a “compressor” for Convolutional Neural Networks (CNN), the OctConv method saves computational resources while boosting effectiveness.
Interactive movies are redefining cinema and storytelling and opening up a world of possibilities in the entertainment industry. There are no “spoilers” for films with no predetermined endings, whose characters and plots develop based on viewers’ real-time direction. Now, what if these viewers became characters?
Chinese technology giant Tencent has open-sourced its face detection algorithm DSFD (Dual Shot Face Detector). The related paper DSFD: Dual Shot Face Detector achieves state-of-the-art performance on WIDER FACE and FDDB dataset benchmarks, and has been accepted by top computer vision conference CVPR 2019.
A collaboration between researchers from China’s Beihang University and Microsoft Research Asia has produced TableBank, a new image-based dataset for table detection and recognition built with novel weak supervision from Word and Latex documents on the Internet.
Artificial intelligence and machine learning have become powerful tools for tech giants looking to build the optimal cloud service platform. To catch up with rivals Amazon Web Services and Microsoft Azure, Google is accelerating the development of its own cloud services.
As robots take over industrial manufacturing, specific and accurate robot control is becoming more important. Conventional feedback control methods can effectively solve various types of robot control problems by capturing structures with explicit models such as motion equations.
Facebook AI Research has announced it is open-sourcing PyTorch-BigGraph (PBG), a tool that can easily process and produce graph embeddings for extremely large graphs. PBG can also process multi-relation graph embeddings where a model is too large to fit in memory.
DeepMind trained and tested its neural model by first collecting a dataset consisting of different types of mathematics problems. Rather than crowd-sourcing, they synthesized the dataset to generate a larger number of training examples, control the difficulty level and reduce training time.
Google Brain researchers have proposed LAMB (Layer-wise Adaptive Moments optimizer for Batch training), a new optimizer which reduces training time for its NLP training model BERT (Bidirectional Encoder Representations from Transformers) from three days to just 76 minutes.
Chinese AI company iFLYTEK has bested the SQuAD2.0 challenge once again. The model “BERT + DAE + AoA” submitted by the joint iFLYTEK Research and HIT (Harbin Institute of Technology) laboratory HFL outperformed humans on both EM (exact match) and F1-score (fuzzy match) indexes to top the SQuAD2.0 leaderboard.
In a scene that looks like it’s from a sci-fi movie, a YouTube video posted today by robotics company Boston Dynamics shows a huge, ostrich-like robot “Handle” whirling round while deftly moving boxes in a warehouse. The video has garnered over 138,000 views in less than four hours.
Andrew Brock, first author of the high-profile research paper Large Scale GAN Training for High Fidelity Natural Image Synthesis (aka “BigGAN”), has posted a GitHub repository of an unofficial PyTorch BigGAN implementation that requires only 4-8 GPUs to train the model.
Facing the incomplete information environment, the asynchronous neural virtual self-play (ANFSP) method allows AI to learn to generate optimal decisions in multiple virtual environments. The approach has performed well in Texas Hold’em and multiplayer FPS video games.
Baidu has released ERNIE (Enhanced Representation through kNowledge IntEgration), a new knowledge integration language representation model which outperforms Google’s state-of-the-art BERT (Bidirectional Encoder Representations from Transformers) in Chinese language tasks.
NVIDIA CEO and Co-Founder Jensen Huang says a rumored next-generation GPU architecture is not a priority for the company, and that he remains optimistic about clearing the chip inventory built up for cryptocurrency mining. Huang made the remarks in a press conference Tuesday at the GPU Technology Conference (GTC) in Santa Clara.
It is no secret that deep neural networks (DNNs) can achieve state-of-the-art performance in a wide range of complicated tasks. DNN models such as BigGAN, BERT, and GPT 2.0 have proved the high potential of deep learning. Deploying DNNs on mobile devices, consumer devices, drones and vehicles however remains a bottleneck for researchers.
DeepMind’s Research Platform Team has open-sourced TF-Replicator, a framework that enables researchers without previous experience with the distributed system to deploy their TensorFlow models on GPUs and Cloud TPUs. The move aims to strengthen AI research and development.