While there is a plethora of innovative artificial intelligence based companies, none come close to Baidu

Baidu Inc., is one of the largest providers of Chinese language Internet services. Today, it is also one of the leading artificial intelligence innovators in the world. The company has helped China position itself on the global tech map while also boosting its economy along with Alibaba and Tencent.

Sources reveal that in 2020 alone, Baidu’s core R&D expenditure accounted for 21.4% of its revenue, becoming one of the top Internet companies with the highest R&D spending. Further, Baidu also claims to have most artificial intelligence-related patent applications in China. This is a testament to Baidu’s long-term commitment to driving technological advancement. Today, Baidu is actively and often successfully integrating artificial intelligence technologies into all of its major businesses. This ranges from search engine, to drug discovery and even autonomous driving. In 2018, Baidu became the first Chinese company to join an artificial intelligence ethics group (Partnership on AI (PAI)) led by top U.S. tech firms, Alphabet Inc’s Google, Apple Inc and Facebook Inc.

Here are some notable innovations in artificial intelligence applications from Baidu:

The COVID-19 Conundrum

During the COVID-19 outbreak too, the company had leveraged its expertise in artificial intelligence, and associated technologies and products, to support frontline efforts to prevent and control the pandemic. It created an artificial intelligence system that uses infrared technology to predict passengers’ temperatures at Beijing’s Qinghe Railway Station. Its Smart Consulting Assistant has also proved resourceful in helping doctors make rapid diagnoses and initiate treatment online.

Last year, Baidu had also open-sourced its Ribonucleic acid (RNA) prediction algorithm LinearFold. This artificial intelligence algorithm aims to accelerate the prediction time of a virus’s RNA secondary structure, which is crucial to understand it and developing vaccines. Researchers found that LinearFold is capable of predicting the secondary structure of the SARS-CoV-2 RNA sequence in only 27 seconds, 120 times faster than other methods. Apart from LinearFold, Baidu has also launched PaddleHelix, a machine learning-based bio-computing framework aimed at facilitating the development of vaccine design, drug discovery, and precision medicine.

Baidu’s Autonomous Vehicles Ventures

 Apollo is an ambitious, open-source platform from Baidu that is designed to support self-driving vehicles. Apollo’s deep-learning inference support is designed to handle complex driving environments, including sensor fusion and AI processing. Baidu’s Automated Valet Parking (AVP) which runs on ACU-Advanced and Xilinx’s hardware is also built on Apollo. Last year, Baidu made headlines for its demonstration of Fully Automated Driving without a safety driver via live streaming. Using Apollo’s new Fully Automated Driving capability, the artificial intelligence system can independently drive without a safety driver inside the vehicle, a breakthrough that will accelerate the large-scale deployment of autonomous driving technology across China.

PaddlePaddle

Baidu’s PaddlePaddle offers software developers of all skill levels the tools, services, and resources they need to rapidly adopt and implement deep learning at scale. It also hosts toolkits for cutting-edge research purposes, like Paddle Quantum for quantum-computing models and Paddle Graph Learning for graph-learning models. Companies like LinkingMed have used PaddlePaddle to develop an AI-powered pneumonia screening and the lesion-detection system being used in the hospital affiliated with Xiangnan University in Hunan Province. By using Paddle Detection, a PaddlePaddle toolkit for image processing, Jinlu Technology trains an instance-segmentation model for sorting waste plastic bottles.

Natural Language Comprehension

One of the most sought after yet trickiest challenges of artificial intelligence algorithms is to enhance its NLP abilities. Baidu’s ERNIE (short for Enhanced Representation through kNowledge Integration), is presently the best in the world by GLUE (General Language Understanding Evaluation) score. ERNIE can understand blocks of language in context and therefore comprehend commands and interactions of all kinds efficiently. Some of the ERNIE’s iterations like ERNIE-GEN enable language generation tasks, like dialogue engagement, question generation, and abstractive summarization. In contrast, ERNIE-ViL helps with visual understanding.

Other Exemplary Inventions

In 2015, Baidu launched its intelligent personal assistant, Duer. Also dubbed as the Chinese Apple Siri, Duer includes multi-modal interaction, natural language processing, and other such technologies for a natural interaction and smarter understanding.

Baidu’s Deep Speech is a state-of-the-art speech recognition system developed using end-to-end deep learning by Baidu Research. It has also developed a production-quality text-to-speech (TTS) system using deep neural network – Deep Voice. Baidu mentions that its Deep Voice is faster and more efficient than Google’s WaveNet. Baidu also has SwiftScribe,  an AI-Powered Transcription Software among its wide array of artificial intelligence innovations. Based on Deep Speech 2, the main function of SwiftScribe is to transcribe audio material into the text in order to solve the problem of consuming a large amount of time-by-word dictation.

Baidu Brain, is another core artificial intelligence innovation from the Beijing based company that features advanced technology for recognizing and processing speech, images and words as well as building user profiles based on big data analysis.

Moreover, last year, Baidu launched its own artificial intelligence-based accelerator called Kunlun K200 SoC. This 256-TOPS accelerator was designed to handle its internal deep-learning workloads. The K200 accelerates common neural-network and SQL operations. On artificial intelligence inference benchmarks, it matches the power efficiency of Nvidia’s T4 card. This year, Baidu will start mass production of Kunlun 2.

The post BAIDU : TOP ARTIFICIAL INTELLIGENCE INNOVATIONS FROM THE CHINESE ‘GOOGLE’ appeared first on Artificial Intelligence.

New Delhi, Sep 16 (PTI) Congress members in Rajya Sabha on Wednesday expressed concern over a media report that suggested tracking of over 10,000 prominent Indian individuals and organisations by a Chinese technology company and asked the government whether it has taken note of the matter.

After the issue was raised by two Congress members, K C Venugopal and Rajiv Satav, during the Zero Hour, Rajya Sabha Chairman M Venkaiah Naidu asked the Minister of Parliamentary Affairs to take note of it and inform the minister concerned.

“I would like to invite the attention of this House through you Chairman, a shocking news which is related to national security and privacy of Indian citizens,” Venugopal said.

He said the Indian Express has reported that a Shenzen-based technology company with links to the Chinese government, and the Chinese Communist Party, is tracking over 10,000 Indian individuals and organisations in its global data base of foreign targets.

“This includes…very shocking, the President of India also you the Vice President of India, the Prime Minister of India and the opposition leaders, including Congress President, Chief Ministers, MPs, army chief, and industrialists,” Venugopal said.

The Chinese company has also collected the data base of bureaucrats in key positions, judges, scientists, academicians, journalists, actors, sportspersons, religious figures and activists, he added.

“This is a major area of concern. I would like to know from the government whether it has taken note of it. If so, what action has been taken,” Venugopal asked.

The Congress leader also charged that the government was not answering and discussing reports regarding Chinese intrusion in border areas.

Echoing his views, Satav said this is a very serious issue.

“The government should give clarification. Truth should come out. How can a Chinese company spy on prominent people,” Satav said.

To this, Naidu asked the Minister of Parliamentary Affairs to take note of the matter.

“It has appeared prominently. So inform the minister concerned and see what can be done and find out the veracity also,” the Rajya Sabha Chairman said. PTI MJH ANZ MJH DV DV

The post Cong members raise in RS reports of data mining by Chinese firm appeared first on Artificial Intelligence.

Beijing based AI startup Megvii announced today that it will open source its deep learning framework MegEngine. The open sourced framework will allow global developers to create AI solutions for industrial and commercial scenarios and foster an ecosystem around Megvii’s new generation AI productivity platform, Brain++.

Initially developed in 2014 and now in its eighth edition, MegEngine is one of the three primary components of Brain++ architecture that Megvii built to train computer vision algorithms at scale. The other two proprietary components of Brain++ include MegData (a data management system) and MegCompute (a computing power dispatching system).

MegEngine has been deployed across Megvii’s three business units (personal IoT, city IoT, and supply chain IoT) and is used by its 1,400+ strong R&D team in its day-to-day algorithm and product development. According to Megvii, MegEngine is especially effective for training large volumes of images and videos and completing complex tasks such as image classification, object detection, object/scene segmentation and video analytics.

MegEngine will be released under the terms of Apache License 2.0, a permissive free software license written by the Apache Software Foundation. Users will be able to distribute, modify, and distribute modified versions of the software without concern for royalties. The MegEngine code will be available on two code repository websites, GitHub and OpenI.

“China has been participating in open source since the Android system was introduced in 2011, and Chinese developers today make up an increasingly large proportion of the global open source developer community,” said Tang Wenbin, Co-founder and CTO of Megvii, “We think that open sourcing MegEngine is consistent with this trend. Megvii wants to lead the way in China and create good will among developers here and abroad.”

Founded in Beijing in 2011 by three Tsinghua University graduates, Megvii is a rising image recognition and deep learning software firm. Dedicated to becoming the first publicly listed Chinese AI company, Megvii plans to refile for a Hong Kong IPO after its first application lapsed in February 2020, according to Reuters. Megvii was blacklisted in the US along with several other Chinese AI companies over alleged human rights violations.

The post Chinese AI Startup Megvii Open Sources Its Proprietary Deep Learning Framework MegEngine appeared first on Artificial Intelligence.

Kennesaw State Professor of computer science Dr. Dan Lo and his team of students created a program last semester to data mine Chinese social media sites in partnership with the U.S. Embassy in Beijing, China.

Lo said the program retrieves and deciphers posts on popular Chinese social media outlets like Weibo and WeChat. Lo’s team of student researchers — four undergraduates and two graduates — conducted extensive data mining on China’s Weibo platform and in turn, gained valuable real-world experience under his tutelage.

Weibo, meaning “micro blog” in Chinese, is an open social platform with over 445 million monthly active users. It is similar to both Facebook and Twitter, and likewise is an epicenter of Chinese news.

Lo said he is passionate about the project as it embodies elements of social networking, big data, machine learning and data science. Additionally, it required fluency in both Chinese and Python, a coding language. The team’s biggest challenge was processing Chinese articles correctly because of the complexity of the Chinese language.

“Chinese language is highly context-sensitive. Consecutive Chinese letters can be combined and interpreted in multiple ways.” Lo said. “For instance, our data mining programs needed to determine if ABC was meant to be read as A, B, C, AB, BC or ABC in the mind of the writer. The meaning in each instance varied vastly.”

Handling word segmentation, numbers, acronyms, synonyms and newly developed words in Chinese slang posed additional frustrations. Linguistic challenges and the sheer volume of the big data analyzed required painstaking efforts to codify the inputs.

“Social media mining in English has been performed for many years in Twitter, Facebook, Instagram and other networks,” Lo said. “While it is theoretically possible to do the same in Chinese by applying most of the English techniques, it is not that simple.”

Information sharing on social media is a valuable tool in forming public opinion. The results from the project enabled Lo’s team to show the information shared among Chinese users and the key statistics of that data.

“Dr. Lo is an excellent mentor because he allows his students to have a role in all stages of the research process,” graduate research assistant Charles Gardner said. “By working so closely with his students, Lo creates strong researchers as well as a great finished project.”

Gardner has been working as a graduate research assistant for Lo for over a year now.

“In English, we do not have word segmentation,” Gardner said. “However, in this project, we look at Chinese words within references and it ends up being an interesting system.”

Lo has already begun work on his next exciting project identifying fake news and dealing with misinformation. This new project has been commissioned by the U.S. Embassy in Poland.

The program was awarded to KSU by the The Diplomacy Lab, an alliance between the U.S. Department of State and U.S. colleges and universities, according to KSU News. Under the program, partner U.S. universities are invited to submit proposals for State Department projects in various humanitarian areas.

The post Computer science team creates coding program to interpret Chinese social media texts appeared first on Artificial Intelligence.