Source  
Deep Learning prefer the human voice -- Just like us! On April 07, Wednesday, 2021

A new study from Columbia Engineering proves that artificial intelligence systems might reach higher levels of performance if they are programmed with sound files of human language. When used to convey information, the language of binary numbers is compact and precise. In contrast, spoken human language is more tonal and analog, and, when captured in a digital file, non-binary. A neural network whose "training labels" consisted of sound files performed better in identifying objects in images than one that used simple binary labels.

Read more at techxplore.com

Source  
Novel Deep Learning Model to predict pharmaceutical properties - MoIMapNet On March 30, Tuesday, 2021

MolMapNet is a new artificial intelligence tool that can predict the pharmaceutical properties of drugs. It can be used by people with little or no knowledge of computer science, biology or other sciences. Researchers found that it outperformed state-of-the-art AI tools on most of the 26 pharmaceutical benchmark datasets. This innovative tool does not require parameter fine tuning, which means it is also accessible to non-expert users, the researchers say.

Read more at techxplore.com

Source  
Artificial Intelligence to improve mobile mixed reality generation On March 25, Thursday, 2021

Japanese researchers have developed a system that can create a mobile mixed reality landscape. They used deep learning to remove obstructions and trees using a video game engine. The work may lead to a revolution in green architecture and city revitalisation. The team hopes this research will help stakeholders understand the importance of GVI on urban planning.

Read more at techxplore.com

Source  
Novel learning framework for symbolic regression On March 22, Monday, 2021

Computer scientists at Lawrence Livermore National Laboratory develop tool that leverages deep reinforcement learning. The tool outperforms baseline methods on benchmark problems. The paper was accepted as an oral presentation at the ICLR 2021 machine learning conference. Symbolic regression aims to uncover the underlying equations or dynamics of a physical process. LLNL's deep learning approach is theory-backed and based on gradient information, co-authors said.

Read more at techxplore.com

Source  
Learning technique arrange sensors on robot's body for efficient operation On March 22, Monday, 2021

MIT researchers have developed an algorithm to help engineers design soft robots. The algorithm suggests an optimised placement of sensors within the robot's body. Rigid robots have a finite array of joints and limbs while soft robots can deform in any way possible. The advance is a step toward the automation of robot design, the researchers say. The research will be presented during April's IEEE International Conference on Soft Robotics.

Read more at news.mit.edu

Source  
Generalization in Deep Learning from a wider viewpoint. On March 11, Thursday, 2021

One of the major mystery in deep learning is the thorough understanding of the concept of generalization. The Deep Bootstrap Framework, a new framework describes a model that has access to an infinite stream of samples, and can be iteratively updated. It was found that models that train quickly on infinite data are the same models that generalize well if they are instead trained on finite data. This connection brings new perspectives on design choices in practice, and lays a roadmap for understanding generalization from a theoretical perspective.

Read more at ai.googleblog.com

Source  
Non-Trivial task in Computer Vision On March 08, Monday, 2021

Non-rigid point set registration is the process of finding a spatial transformation that aligns two shapes. It has extensive applications in areas such as autonomous driving, medical imaging, and robotic manipulation. A researcher from Kanazawa University has demonstrated a technique that reduces the computing time. The results demonstrate that the proposed technique is efficient even for point sets with more than 10 million points, the researchers say.

Read more at techxplore.com

Source  
Learnings from the top AI summit : RE-WORK Deep Learning 2.0 On March 01, Monday, 2021

January end witnessed one of the most renowned virtual summit of deep learning, RE-WORK 2.0. Speakers at this auspicious summit were professors from renowned universities and researchers from big tech firms. Aim was to bridge the gap between the latest technological research advancements and real world applications in business and society.

Read more at medium.com

TAGS