Representations for Language: From Word Embeddings to Sentence Meanings
This talk was recently given by Prof. Christopher Manning at Simons Institute, UC Berkeley. It is an introductory tutorial without complicated algorithms.
Category: Research
Technical review of the newest machine intelligence research.
1 Day Countdown: Industry Spotlights on Global Machine Intelligence Summit 2017
With just one day remaining before the GMIS 2017 kicks off in Beijing, we’ll survey the industry applications of AI technologies.
AlphaGo: The Ultimate Go Master
For more than 2,500 years, humans have been playing and studying the incredibly complex game of Go. In a matter of months, AI learned how to do it better.
3 Day Count Down: GMIS 2017 Warms Up With Debate at Pre-Event in San Francisco
“There is going to be a revolution that’s 100 times the size of the mobile revolution,” claimed Saman Farid fromContinue Reading
Cool IJCNN Stuff from Alaska
From May 14 to 18, the 30th International Joint Conference on Neural Networks (IJCNN 2017) was held in Anchorage, AK, USA. Continuing the long tradition, the conference is organized by the International Neural Network Society (INNS), in cooperation with the IEEE Computational Intelligence Society (IEEE-CIS).
Generative Face Completion
This paper proposes a deep generative model for face completion, which can directly generate facial components for the missing regions of a face image.
Google I/O 2017: Integrating Everything with AI
At the 10th Google I/O, held May 17-19 at the Shoreline Amphitheatre in Mountain View, California, Google took a different approach from unveiling exciting new products, by putting its focus on the convergence of existing products, aimed at providing a better user experience.
Big Picture Machine Learning: Classifying Text with Neural Networks and TensorFlow
TensorFlow is one of the most popular open source AI libraries. Its high in computing efficiency, and the rich development resources make it widely adopted by companies and individual developers.
Face Aging with Conditional Generative Adversarial Networks
The main advantage of the acGAN is the “Identity-Preserving” latent vector optimization approach to maintain the original person’s identity in reconstruction.
1960s-2010s: Humanistic Intelligence and History of Wearable Computing
Personal computers and mobile devices are in their heyday. Researchers are swarming standalone AI, focusing on how to automate self-learning intelligent systems. The interfaces for wearables meanwhile are evolving from smart screens to gesture commands, like those often seen in AR and VR commercials.
Mixtures of Lightweight Deep Convolutional Neural Networks: Application in Agricultural Robotics
Nowadays, the use of agricultural robots are steadily rising, like the AgBot II, which helps detect and classify weeds, and Harvey, used for detecting and segmenting crops.
Why AlphaGo is not AI
Thanks to AlphaGo, the term “Artificial Intelligence” (AI) came under the spotlight (again). However, Jean-Christophe Baillie doesn’t agree that AlphaGo is AI, for the reason that it’s not able to get us to full AI – an Artificial General Intelligence (AGI).
A Guide to Receptive Field Arithmetic for Convolutional Neural Networks
This is a comprehensive guide on how to calculate and visualize the receptive field information of a convolutional neural network.
Automated Synaptic Connectivity Inference For Volume Electron Microscopy
Researchers introduce an automated synaptic connectivity inference pipeline (SyConn) to delivers a richly annotated wiring diagram, or components of a connectome.
Understanding CNN from Signal Perspective
RECOS model supply us with a signal analysis perspective of Convolution Neural Network. From this perspective, we can see the effectiveness of activation and deep architectures.
Unsupervised Image-to-Image Translation with Generative Adversarial Networks
This paper proposes a general approach to achieve “Image-to-Image Translation” by using deep convolutional and conditional generative adversarial networks (GANs).
Stochastic Neural Networks for Hierarchical Reinforcement Learning
Authors proposed a general framework for learning a span of skills in a pre-training environment, which can be used in downstream tasks by training a high-level policy on top of the skills.
Large-Scale Evolution of Image Classifiers
This work proposes a novel neural-evolution algorithm for large scale search.
How to Train a Very Large and Deep Model on One GPU?
The article discusses about the limitation of using existing GPU to train neural network and how one can improve to train larger model.
The Time to Marry AI May Come Soon
Dr. Eisaku Maeda, the director of NTT Communication Science Laboratories, was recently interviewed by The Huffington Post, sharing his opinions on AI.
Can AI Fool Humans? In the Spirit of April (Fools): This is No Joke
In the spirit of April (Fools), let us see how far down is AI along the path of “fooling” humans.
A Brief Introduction to Humanistic Intelligence: The Symbiotic Future of Machine and Human
“Why did I go to such extremes? Because I realized that the future of computing was as much about communications between people wearing computers as it was about preforming colossal calculations.” — Steve Mann
Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network
This paper proposed a computationally efficient convolutional layer to upscale the final low-resolution feature map to a high-resolution output.
What Does It Take For Intel To Seize The AI Market?
In this article we will provide some insights of Intel’s recent deep learning products.
A Robust Adaptive Stochastic Gradient Method for Deep Learning
This paper proposes an adaptive learning rate algorithm, which utilizes stochastic curvature information of the loss function for automatically tuning the learning rates.
Neural Networks for Beginners
This review will introduce and compare the three commonly used machine learning frameworks: Matlab Neural Network Toolbox, Torch and Tensorflow.
Universal Schema: Bridging the Gap Between Strings and Things
Andrew McCallum explained how the use of Universal schema can improve knowledge representation and reasoning from natural language.
The Future of Computing: Neuromorphic
Neuromorphic computing, a promising direction for future of computing that emulates human brain to maintain energy efficiency and fast computing speed.
Alibaba’s New Study on Artificial Intelligence: Achieving Multi-arms Cooperative Gaming in “StarCraft”
Researchers from Alibaba and UCL have recently published a new study, claiming that multiple game units controlled by artificial intelligence are now able to develop tactics similar to human players using unsupervised learning.
Synced Exclusive Interview with “Fine Art” Team Lead: “Stronger Models though Brand New Reinforcement Learning Techniques”
Synced had the opportunity for an exclusive interview with Tecent AI Lab’s senior manager, team lead for Fine Art, Liu Yongsheng.
Canada’s Future Bets on AI
Canada unveiled a budget of $125 million to launch Pan-Canadian Al Strategy, a new independent institute for AI, Vector Institute.
Pixel Recursive Super Resolution
Google released a method for recovering pictures from low resolution to high resolution using pixel recursive super resolution model.
Hey Computer, What If I Don’t Want to Solve my Schrödinger Equation?
Compared to simplifying or approximating the solution of a complicated equation, deep learning provides a highly flexible model and might serve as a perfect walk-around method.
How Artists Can Use Neural Networks to Make Art
David Aslan uses neural network “Deep Style” to transform original photos or paintings into images with other styles.
PrOCTOR: Predicting Failure Probability of Clinical Trials for Potential Drugs
PrOCTOR integrates chemical properties of a compound and its protein targets, to provide a measure about whether the drugs in clinical trials will fail for toxicity reasons.
Learning to Reinforcement Learn
The authors introduced a novel approach called deep meta-reinforcement learning (meta-RL) which can rapidly adapt to new tasks.
ArtGAN – Artwork Synthesis with Conditional Categorical GANs
ArtGAN can generate images with abstract information like images with a certain art style.
Optimizing Neural Networks using Structured Probabilistic Models
This talk presents natural gradient, a new approach to approximate the Fisher matrix by adopting structured probabilistic models.
Why AI Must Be Biased, and How We Can Respond
AI must be biased because the knowledge we used in the training process contains traces of our history, including our prejudices.
Fei-fei Li in Google Cloud NEXT ’17: Annoucing Google Could Video Intelligence API, and more Cloud Machine Learning Updates
Between March 8-10, Synced was invited as media guest to attend the Google Cloud NEXT ’17 conference in San Francisco.