"New Artificial Intelligence Inspired by Human Brain Function"

2020-02-02

Recently developed at the University of Liège, an algorithm based on neural modulation enables intelligent agents to adapt quickly to new tasks, revolutionizing AI technology. Published in PLOS ONE, this groundbreaking research combines neuroscience and AI to enhance learning capabilities.

Read more

Combination of Underlying Foundations and Upper-layer Applications

2020-02-03

2015 was a pivotal year for China's IT industry with enterprise dataization gaining traction. Entrepreneurs like Sang Wenfeng of Sensors Data aim to address data collection and analysis gaps. Their tool Sensors Analytics offers unique cloud deployment options and a robust data foundation for Chinese enterprises.

Read more

Uber's Comprehensive Training Data Boosts Deep Learning Speed by 9 Times

2020-01-22

Uber AI Lab developed Generative Teaching Networks (GTN) to create training data for neural networks, accelerating the training process by 9 times. GTN-NAS outperforms other NAS methods by using less computational resources, allowing quick model evaluation. Uber's system uses meta-learning to generate training samples, achieving high accuracy with fewer training steps.

Read more

Facebook AI Achieves 107x Faster Virtual Agent Training速度

2020-01-22

Facebook AI introduces DD-PPO, an RL algorithm for target navigation without maps. Achieved 99.9% success in various virtual environments, increasing AI application potential. Proposed scalable RL approach for efficient training. Trained agents with DD-PPO showed near-perfect target navigation. Future focus on enhancing machine learning in real-world navigation tasks. Exciting potential to expand DD-PPO to more complex navigation challenges. Repository link: https://github.com/facebookresearch/h

Read more

2020 Robot Industry New Trends Forecast and Outlook

2020-01-22

The industrial robot market in China faced a downturn in 2018 but is recovering in 2019, with robots proving value in addressing labor shortages and undertaking dangerous tasks. In 2020, collaboration among robot companies, acquisitions by large firms, and expansion of drone deployment are expected. The 2020 Olympics in Japan will showcase robotics technology, while autonomous vehicles will focus on controlled environments. Retailers will use robots to increase capacity, and mobile manipulation

Read more

"Development Trend of the Next Decade in Lanzhou New Area's Construction of Smart City"

2020-01-22

In the next decade, 70% of China's population will live in cities, generating 80% of national income. Lanzhou New Area is actively building a smart city with big data, IoT, AI, and blockchain industries, creating safer and more efficient urban services. The future of smart cities will focus on utilizing ICT technologies, investing in IoT, data interoperability, and urban communication infrastructure. As the digital economy era progresses, smart city development will continue to expand, with 5G p

Read more

Baidu Brain Intelligent Creation Platform Makes Beautiful Videos Even More Beautiful

2020-01-23

Value Achievements: Haokan Video integrates Baidu Brain AI for automated video production, saving 5 million yuan annually. It improves timely resource video coverage. Core Demands: Haokan Video aims to provide high-quality, timely short video content to meet increasing demand. Solution: Leveraging Baidu Brain AI, Haokan Video quickly generates videos based on article links, reducing production costs.

Read more

Researchers Use AI for Deblurring Faces in Photos

2020-01-23

A joint effort by researchers from the UAE, Beijing, and Stony Brook produced an AI system for post-production deblurring, focusing on faces. The model, trained on a diverse dataset, boasts state-of-the-art performance in removing blur from dynamic images, outperforming existing methods. This user-friendly network excels in capturing clear structures and semantic details by incorporating features from various domains, providing superior restoration effects for both foreground and background elem

Read more

Google Releases Most Detailed 3D Brain Scan Image in History

2020-01-23

Google and its partners released the largest brain scan collection ever, mapping 20 million synapses and 25,000 neurons in fruit fly brains. This detailed "hemibrain" connectome is a significant advancement in neuroscience research, providing a public resource for scientists to study brain connectivity.

Read more

Adobe's AI Virtual Fitting Room Launched, Say Goodbye to Awkward Seller Shows

2020-01-23

In the past decade, virtual try-on technology has gained popularity. Adobe, IIT, and Stanford developed SieveNet, achieving state-of-the-art results in image-based virtual try-on. It can map clothing onto virtual objects accurately, handling occlusions, pose variations, and improving overall quality preservation. SieveNet can be easily integrated into existing platforms, enhancing online fashion transactions.

Read more

"World in the Eyes of Computers: Breaking Free from Two Dimensions"

2020-01-23

Image recognition, driven by AI, is evolving from traditional deep learning to geometric deep learning, inspiring breakthroughs in physics. Equivariantly equivariant CNNs now recognize patterns on spherical surfaces. This rethinking of neural networks pushes AI beyond 2D limitations, addressing the issue of data efficiency. The journey to advancement continues.

Read more

Intel Utilizes Computer Vision to Help Penguins Avoid Extinction

2020-01-21

The population of Antarctic emperor penguins is at risk of extinction by 2100 due to climate change. Intel's high-tech group developed a computer vision solution to help ecologists identify penguins faster and more accurately. Gramener used AI to estimate penguin quantities from images, reducing training time significantly. Other companies like Google and Microsoft are also using AI for ecological research, such as tracking African elephants and wildlife in the Serengeti. Intel's TrailGuard AI s

Read more

Beyond Machine Learning to Machine Reasoning

2020-01-21

Conversations on AI encompass various topics like machine learning, autonomy, and research history. The cycle of AI interest, investment, hype, and decline defies the typical technology life cycle. Understanding AI and machine reasoning pave the way for overcoming current limitations.

Read more

Where Have All the AI Startups Gone? A Review of the 2019 Acquisitions by Tech Industry Giants

2020-01-21

The battle for AI talent intensifies as tech giants like Google and Apple acquire AI startups. Shortage of skilled professionals persists, with high demand in robotics, computer vision, and natural language processing. Large tech companies ramp up acquisitions to stay competitive in AI. Acquisition examples include Facebook's GrokStyle for visual search, Ctrl-labs for brain-computer interface, and Servicefriend for AI chatbots. Amazon's acquisitions include TSO Logic for cloud spending analysis

Read more

Legislation on Artificial Intelligence: The Inevitability and Initiation of Industry Standardization

2020-01-21

Artificial intelligence challenges traditional ethics with incidents like AI-generated fake news and concerns over safety and privacy. Legislation is crucial for AI industry development, with countries like China and the EU leading the way. AI governance globally remains a significant challenge, with potential negative impacts on social fairness and inequality. AI legislation must balance economic significance with social norms to ensure human values are prioritized.

Read more