Shenzhen Based RoboSense is the Next DJI for LIDAR
The global market for this versatile, high-precision, long-distance measuring “eye” will reach US$36 billion in 2030.
Author: Synced
Machine Intelligence | Technology & Industry | Information & Analysis
A Brief Overview of Attention Mechanism
Attention is simply a vector, often the outputs of dense layer using softmax function.
Machine Teaching A New Paradigm for Building Machine Learning Systems
This paper laid out the position of Machine Teaching, a universally accessible, easy to manage, and fast discipline proposed by Microsoft Research, and articulated the fundamental machine teaching principles.
Baidu All in AI: DuerOS and Apollo Open Platforms
The “Baidu AI Developers Conference” (Baidu Create 2017) was held July 5 at the China National Convention Center in Beijing.Continue Reading
Best Time Ever for AI Startups?
While tech giants like Google and Amazon are setting the pace in AI R&D these days, startups can also find success by targeting niche markets
Adversarial Training Methods For Semi-Supervised Text Classification
In applying the adversarial training, this paper adopts distributed word representation, or word embedding, as the input, rather than the traditional one-hot representation. The reason lies in the fact that the higher dimensionality the input has, the more likely it is to be disturbed by noise.
Bridging Neural Machine Translation and Bilingual Dictionaries
In this paper, the author aims to design a smart way to handle rare/unknown words in the bilingual training data during machine translation.
What AI Talent Are Companies Hungry For?
As more companies bet on AI, computer scientists with AI expertise are coming into demand on the job market.
AI Biweekly: 10 Bits from Sept (Pt 1)
09/05 — Google Updates Its Street View Cameras with AI and Machine Learning Google just upgraded its street view camera for the firstContinue Reading
Wearables: Translate Ideas into Products
Notable wearables today include Fitbit wristbands, the Apple watch, Google Glass, and Playstation VR — all of which have gained great market shares. As the data-collecting portal to the IoT, wearables hold many future possibilities.
English-Japanese Neural Machine Translation with Encoder-Decoder-Reconstructor
This paper is the implementation of ‘encoder-decoder reconstructor framework’ for neural machine translation for the English-Japanese translation task.
Is AI Advancing Too Quickly or Too Slowly?
Last month, it was widely reported that two Facebook chatbots had been shut down after communicating with each other in an unrecognizable language.
Call to Participate! Alibaba’s Ant Financial Hosts First Hackathon in San Francisco
This October, Ant Financial and Synced are co-hosting the ATEC Developer Creativity Challenge for developers to showcase creativity using Alipay Mini-program.
Apple Embraces AI with iPhone X
The latest iPhone struts its AI strength with three new features: Face ID, for facial recognition; Animoji, an emoji animator; and a new camera exposure mode called portrait lighting.
AutoX Wants to Put a Self-driving Car in Your Driveway in Two Years
Jianxiong Xiao (肖健雄), founder of startup AutoX, developed his first self-driving car in six months with seven cameras costing a total of just US$500.
NASA is Using Intel’s Deep Learning to Find Better Landing Sites on the Moon
Recently NASA announced that their Frontier Development Laboratory (FDL) would cooperate with Intel to apply deep learning for space exploration.
Language Model: A Survey of the State-of-the-Art Technology
In this paper, we present a survey on language models, which mainly consists of count-based and continuous-space language models.
A Brief Review of FlowNet
Recently, CNNs have been successfully used in estimating optical flow. Compared with traditional methods, these methods achieved a large improvement in quality. Here, we will give a brief review on the following papers.
Pittsburgh’s Pivot to Artificial Intelligence
Synced meets the people transforming a former steel town into a metropolis powered by artificial intelligence and robotics.
InfiniteBoosting – A Bagging-Boosting Hybrid Algorithm
InfiniteBoosting overcomes the disadvantages of both bagging and boosting, thus slightly improves the performance. Its most important aspect is its adaptability.
AI Biweekly: 10 Bits from August (Pt 2)
Google, Apple, Walmart, Intel, Baidu, Alibaba, Siemens, Salesforce, IBM: August digest from industry giants
RACECAR: A Powerful Platform for Robotics Research and Teaching
RACECAR is a Powerful Platform for Robotics Research and Teaching. It is one of the greatest university courses in this field.
From Amazon to Alibaba: The Quest of Computer Vision Scientist Xiaofeng Ren
“I am tempted by Alibaba’s large pool of business scenarios and huge user base,” says Xiaofeng Ren.
Huawei Announces Kirin 970: AI In Your Phone
China’s biggest domestic mobile phone maker, Huawei, is now powering its phones with artificial intelligence (AI) engines. The Chinese techContinue Reading
System-Aware Distributed Optimization for Machine Learning
The goal of this report is to inspire our readers in the field of distributed machine learning and invite more to join our discussion to exchange knowledge and contribute to our intellectual community.
Technology Requirements for Deep and Machine Learning
The blog author, Rob Farber, has been working on machine learning and related fields as a staff scientist, and participating in relating projects since the 1980s.
GoodAI – Sending AI to School
Marek Rosa’s dream is to create a machine that improves itself recursively to achieve an intelligence level humans cannot imagine, allowing it to tackle problems of any scale.
How Deep Learning May Transform Location-Aware Computing
This article reviews two recent attempts on introducing deep learning models into location-aware computing, effectively reducing expert involvement.
Experts Warn Robots Are Growing in Consciousness and Should Be Classified as an “Invasive Species”
Some experts from different fields stated their concerns that with the rapid rise of artificial intelligence in the next few decades, intelligent robots would take over the human world and extinguish the entire human culture.
Learning to Guide Task and Motion Planning Using Score-Space Representation
This paper proposes a learning algorithm to speed up search in Task and Motion Planning problems (TAMP), and to solve three challenges in order to improve planning efficiency.
A Self-Driving Wheelchair
Autonomous wheelchair developed by researchers at MIT and in Singapore promises improved independence for the disabled Abstract After building self-drivingContinue Reading
Memory, Attention, Sequences
In the article “Memory, attention, sequences”, the author predicts that future work on neural networks will emphasize understanding complex spatio-temporal data from the real world, which is highly contextual and noisy.
How to Control Those Extra Robotic Limbs You’ve Always Wanted
Recently, at an IEEE Conference, MIT researchers proposed a new control method which provides easier and more stable feedback that does not necessarily require existing limbs to control the artificial limbs.
ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices
The ShuffleNet utilizes pointwise group convolution and channel shuffle to reduce computation cost while maintaining accuracy. It manages to obtain lower top-1 error than the MobileNet system on ImageNet classification, and achieves ~13x actual speedup over AlexNet while maintaining comparable accuracy.
A $30 AR Headset?
Amber Garage’s HoloKit is quickly positioning itself as a market stimulator. Just as Google Cardboard increased access and interest in virtual reality, so might HoloKit for augmented reality.
CVPR2017 Honorable Mention Paper – Annotating Object Instances with a Polygon-RNN
This paper proposes not only a method as a useful tool but also a new concept for modeling complex annotation as a simple polygon.
Deep Learning in Real Time – Inference Acceleration and Continuous Training
In this report, we will touch on some of the recent technologies, trends and studies on deep neural network inference acceleration and continuous training in the context of production systems.
Accelerating Convolutional Neural Networks on Raspberry Pi
This method, with considerable computation time and controllable model size, is particularly simple and very useful for implementations on mobile devices such as phones, drones, etc.
IJCAI 2017: Tencent X Synced Young Scholar Tech Night and Reception in Melbourne
International Joint Conference on Artificial Intelligence (IJCAI) 2017 will be held in Melbourne, Australia from August 19 to August 25, summoning top researchers from around the world.
History and Frontier of the Neural Machine Translation
Compared to SMT, NMT can train multiple features jointly and does not need prior domain knowledge, enabling zero-shot translation. In addition to higher BLEU score and better sentence structure, NMT can also help reduce morphology errors, syntax errors, and word order errors of SMT.