Framework for Avoiding "Adverse Outcomes" in Machine Learning

2019-12-11

Researchers from Stanford University and University of Massachusetts Amherst have developed a framework for designing machine learning algorithms with security and fairness constraints. This shift in responsibility from users to designers aims to prevent undesirable behavior in algorithms. The framework consists of three steps and has been tested on student performance data, showing a reduction in gender discrimination. The researchers believe this framework can open the door for new, safer ML a

Read more

Baidu Releases Next-Generation AI Ecological Panoramic Map

2019-12-11

The Baidu Map Ecological Conference unveiled the "New Generation AI Map" and released the first map developer trend report. Baidu Map has become China's largest intelligent location service platform, supporting various industries and transportation upgrades. With 1.8 million registered developers, Baidu Map offers diverse services, including intelligent travel and tourism experiences. Partnering with top institutions, Baidu aims to drive innovation in transportation and travel.

Read more

"Scene Library Released in Sensors Analytics 1.16 Version, Practical Methodology Integrated into Product, Empowering Multi-role Applications in All Aspects"

2019-12-12

Sensors Data's latest release, Sensors Analytics 1.16, includes the "Scenario Library" for practical methodologies, "Public Overview" for enterprise collaboration, and enhanced "Customized Permissions" system. This upgrade focuses on user-friendly intelligent applications, improving data analysis efficiency and operational experiences.

Read more

Facebook, Microsoft, and other companies launch Deepfake Detection Challenge

2019-12-12

The proliferation of deepfake counterfeits, driven by AI algorithms, poses a growing concern with 14,698 videos in June and July, an 84% increase from last year. Major tech companies and academic institutions join forces to combat this issue at NeurIPS 2019. Facebook, AWS, and Google allocate resources to incentivize participation in the global challenge to develop technology for detecting AI-altered videos. Participants must agree to open-source their work for evaluation and stand a chance to w

Read more

Google Utilizes Artificial Intelligence to Enhance Search Results

2019-12-13

Google announced using AI & ML to enhance search results. Users in the US searching for news topics in English see top articles. Results prioritize relevance and quality, offering diverse perspectives. New story understanding tech refines search structure, clustering results crucial for navigating desired content. BERT improves search intent understanding. Search now includes famous quotes and comments. Updates aim to deepen understanding. New features, like "For You" and News Showcase, recommen

Read more

Microsoft's FastSpeech AI Accelerates Natural Voice Generation

2019-12-13

State-of-the-art text-to-speech models, while producing human-like sound, often suffer from slow inference. Researchers from Microsoft and Zhejiang University have developed FastSpeech, a novel machine learning model that addresses this issue by improving performance and eliminating word repetitions or omissions. FastSpeech includes a length regulator to adjust differences between mel-spectrogram and phoneme sequences, achieving a speech speed increase of 0.5x to 1.5x without sacrificing accurac

Read more

Sarcos Robotics unveils exosuit, effortlessly lifting 200 pounds

2019-12-12

Sarcos Robotics unveiled the Guardian XO industrial exoskeleton, a battery-powered system designed to lift 200 pounds. The exoskeleton features 24 degrees of freedom, improved hardware, user-friendly controls, and safety measures. Delivery to advisory groups and the US military begins in January 2020.CEO Ben Wolff emphasized the potential benefits for industries facing worker shortages and safety concerns.

Read more

Application of AI and Robots in the Medical Industry

2019-12-10

Artificial intelligence improves healthcare by reducing errors, enhancing patient health, and aiding decision-making. AI in drug development boosts efficiency and accuracy, leading to new breakthroughs. AI and robotics have the potential to transform healthcare, with organizations adopting AI-driven technology for decision-making and autonomous systems expected to make critical decisions independently.

Read more

"NVIDIA Develops AI that Can Create 3D Models from 2D Images"

2019-12-10

Nvidia introduces DIB-R, a new concept combining learning-based prediction and interpolation to render 3D objects from 2D images. The neural network can generate textured 3D models in just 100 milliseconds, revolutionizing applications in robotics, autonomous vehicles, and safety systems. Future plans aim to turn DIB-R into a virtual reality renderer for creating immersive 3D worlds from photos.

Read more

After I leave, someone will finally accompany you in my place.

2019-12-10

Cats, like humans, have needs for love and care. If your cat shows signs of depression, consider getting a companion. Ebo Robotics has launched a smart cat robot that acts as a playmate, personal trainer, nanny, and health guardian. It can interact with your cat, monitor its activities, and ensure it gets enough exercise. It is also wireless and can self-charge, providing optimal physical and mental health for your furry friend.

Read more

Stack Data Attends the First Insurance Technology Entrepreneurship Week • Xingchenhui Insurance Entrepreneurship Project Roadshow

2019-12-11

In China, insurance technology startups are booming, driving industry transformation. The first Xingchenhui Insurance Entrepreneurship Project Roadshow aims to promote cross-industry collaboration. Zhanlue Data focuses on health insurance risk management through innovative strategies such as a knowledge base system, medical knowledge graphs, natural language models, and abnormal diagnosis models, enhancing data quality and automation in the insurance industry.

Read more

Deep Learning Facilitates Understanding of Genetic Interactions

2019-12-11

Computer scientists at Carnegie Mellon University use deep learning to analyze gene relationships, transforming data into images for CNN analysis. CNNC method shows potential in identifying disease-related genes and drug targets. The method can be applied to various analyses, with data quantity being a main factor for effectiveness.

Read more