"Google AI Unveils 'Visual Task Adaptation Benchmark'"
2019-11-07
Deep learning revolutionizes computer vision, but training from scratch is data-intensive. VTAB protocol offers a unified benchmark for the evaluation of visual algorithms, focusing on task adaptability with minimal restrictions. Researchers found self-supervised models outperform generative models, and S4L combines supervised and self-supervised training for best results. VTAB's code on GitHub ensures reproducibility and accessibility for industry-wide use. Check out the link for more details.
Read more ›Google Launches App Defense Alliance to Combat Malware Distribution
2019-11-07
Google launched App Defense Alliance with EST, Lookout, and Zimperium to combat malicious Android apps. Partners will integrate scanning engines with Google Play Protect for improved detection. Play Protect reduced harmful apps by 0.08% in Google Play. New alerts aim to inform users of potential threats.
Read more ›"Legendary Retirement: Guido van Rossum and His Python Legacy"
2019-11-01
Guido van Rossum, the creator of Python, recently announced his retirement from Dropbox, leaving a legendary legacy. Python, known for its simplicity and extensibility, has become a popular programming language, especially in the field of artificial intelligence. Its rich library and user-friendly features make it the top choice for many developers and major companies. With Python's widespread adoption and impact on AI development, it continues to thrive even after Guido van Rossum's retirement.
Read more ›MIT's Swarm Robotics Research
2019-11-01
MIT team develops self-assembling robot cubes that communicate via barcode-like system, enabling them to achieve true cellular coordination. These "M-blocks" can move, connect with magnets, and perform tasks. The system is cheap, robust, and scalable, with potential for large-scale applications.
Read more ›"DotData Raises $23 Million for Automating Data Science Tasks"
2019-11-01
DotData, a California-based automation tool manufacturer, raised $23 million in funding. CEO Ryohei Fujimaki plans to use the funds for product development and marketing. DotData, a NEC derivative, provides a powerful machine learning automation solution for Fortune 500 companies. JAFCO partner Kitazawa Tomotake praised DotData's capabilities in automating the data science process. Despite competition, DotData is set to lead in the AutoML market space, with the big data market projected to reach
Read more ›Microsoft's AI Project to Replace Driving Instructors
2019-11-01
A new project named HAMS by Microsoft Research India aims to improve road safety using smartphones as low-cost sensing devices. Cameras track driver behavior without interference, promoting road safety. HAMS is currently undergoing driver training tests in India.
Read more ›Brain-Computer Interface Enters Campus, Now Powerful Enough to Read Images in the Human Brain?
2019-11-01
A brain-computer interface device, like FOCUSONE, is used to monitor attention levels in classrooms. Neurobotics & MIPT created a new algorithm using artificial neural networks to display brain images in real-time. Invasive & non-invasive interfaces are used interchangeably in various fields like healthcare, improving brain disease diagnosis & treatment.
Read more ›MIT Launches Artificial Intelligence Simulating Human Brain Memory Process
2019-11-04
World-famous paintings are memorable due to bright colors, simple backgrounds, and central composition. MIT's study uses machine learning to create memorable images. GANalyze can generate impactful graphics for educational purposes and aid in training automated systems for object recognition.
Read more ›Pennsylvania University Professor Uses Artificial Intelligence to Design New Antibiotics
2019-11-04
Since penicillin's discovery in 1928, researchers struggle with antibiotic resistance. De la Fuente uses AI at UPenn to design new drugs from amino acids, aiming to combat bacterial defenses and create more effective antibiotics like Punicalagin 2.
Read more ›US Sanctions New Measure? GitLab Openly Rejects Chinese, Russian Employees for Some Positions
2019-11-04
GitLab announced it will no longer offer services to Chinese and Russian citizens due to industry practices amidst geopolitical concerns. The decision has sparked debates online. GitLab's CEO confirmed the intentional move despite backlash. This reflects US policies and impacts the tech community. Academic institutions have also voiced support for equal treatment of students and faculty.
(Note: 372 characters, need to be shortened to 200 characters)
Read more ›Driving Instructors, You May Be Facing Unemployment Soon
2019-11-04
Microsoft Research has developed the world's first AI driving instructor, called the HAMS system. It aims to replace human driving instructors and reduce learner drivers' nervousness during traditional driving exams. The system uses sensors and computer vision technology to monitor the driver, driving conditions, and environment. It has been successfully implemented in India and may revolutionize the driving education industry, providing a standardized learning experience and potentially reducin
Read more ›Facebook's Slow Video Classifier AI
2019-11-05
Primates' retinal ganglion cells process visual data at different speeds, inspiring Facebook AI Research's "Slow AI" project. SlowFast framework, based on PyTorch, excels in video recognition, leveraging slow motion to enhance accuracy in action classification. This innovative approach fuses information from fast and slow paths, achieving top performance on DeepMind's Kinetics-400 and Google's AVA datasets.
Read more ›Improving Nuclear Power Plant Outage Scheduling with Artificial Intelligence
2019-10-30
The advancement of AI tech like machine learning & NLP has enabled automation of complex tasks with potential to reduce repetitive work, demonstrated by OPG's nuclear outage planning process, leveraging AI solution to optimize task scheduling and enhance efficiency.
Read more ›Google Cloud AI Platform Update: Enhanced Training and Inference Capabilities
2019-10-30
Google Cloud AI Platform has updated its machine learning capabilities, allowing custom containers for training models and introducing expansion tiers for hardware configuration. The platform also supports hosting and deploying ML models with a range of machine types and GPU options for inference. Customers can now log predictions for analysis and improvement.
Read more ›AI-Powered Tool in Latest Photoshop Enables Rapid Selection of Complex Objects
2019-10-30
Adobe's new AI engine simplifies object selection in Photoshop. The "Object Selection" tool allows for precise and quick isolation of complex objects, saving time and reducing tedious work.
Read more ›"Talking about love? The kind that can go home and harvest wheat..."
2019-10-30
Harvest is coming, but Xiaoli's ex Ahai is tired of doing farm work. Driverless wheat harvesters are taking over in Shandong, using IoT, big data, AI, and satellite positioning. Foreign farmers are also adopting this technology. Smart agriculture is the future!
Read more ›2019 International Conference on Computer Vision: China's Strong Academic Influence
2019-10-30
ICCV held in Seoul, co-chaired by top professors. Chinese scholars excelled with most accepted papers and awards. SenseTime, a leading AI company, showcased groundbreaking research at ICCV 2019, demonstrating remarkable innovation and contributions to the global academic community.
Read more ›IBM: Artificial Intelligence Will Transform Every Job and Increase Demand for Creative Skills
2019-10-31
AI is reshaping tasks between humans and machines, impacting job market. Tasks are decreasing as automation rises, creating new opportunities but emphasizing the need for "soft skills." AI is driving demand for intellectual and analytical abilities, with high-end financial and service-oriented jobs seeing wage increases. The pace of change is slow but will accelerate with AI development, necessitating HR reform and employee training in creative thinking.
Read more ›Estimating Heart Rate from Facial Images using Artificial Intelligence
2019-10-31
Chinese researchers introduce RhythmNet, an AI-based heart rate estimator using facial images. It detects blood volume changes in skin tissue with RGB sensors to predict heartbeats per minute accurately. Research shows promising results with error rates below 5 beats/min. Future plans include improving the model with transfer learning and exploring its potential for other physiological measurements.
Read more ›Google's Live Caption adds real-time subtitles to videos.
2019-10-31
The introduction of Live Caption by Google AI team enables real-time subtitles on Pixel 4, improving accessibility and privacy while reducing latency. This feature, based on advanced speech recognition models, aims to expand language support and enhance subtitle accuracy on all Android devices in the future.
Read more ›