Goto

Collaborating Authors

Sensing and Signal Processing


Government presses ahead with Cummings' data science revolution

#artificialintelligence

A British artificial intelligence firm involved in the Vote Leave campaign has been handed a £400,000 contract to tap data from places such as social media sites to help steer the Government's response to Covid-19. Official documents from the Government show Faculty Science was awarded the contract by the Ministry of Housing, Communities and Local Government (MHCLG) in April to provide data scientists who could set up "alternative data sources (e.g. They would, the contract said, apply data science and machine learning to the data, which could help identify trends, and then develop "interactive dashboards" to inform policymakers. It is understood the contract, awarded through the Government's G-Cloud framework, was designed to address an urgent need for the department to analyse real-time data and monitor the effect of Covid-19 on local communities. Faculty's AI technology can be used to process vast amounts of data and in the past was used for polling analysis by the Vote Leave campaign, run by Boris Johnson's adviser Dominic Cummings.


From Digitized Images to Online Catalogs Data Mining a Sky Survey

AI Magazine

The value of scientific digital-image libraries seldom lies in the pixels of images. For large collections of images, such as those resulting from astronomy sky surveys, the typical useful product is an online database cataloging entries of interest. We focus on the automation of the cataloging effort of a major sky survey and the availability of digital libraries in general. The SKICAT system automates the reduction and analysis of the three terabytes worth of images, expected to contain on the order of 2 billion sky objects. For the primary scientific analysis of these data, it is necessary to detect, measure, and classify every sky object.


Learning Compact Visual Descriptors for Low Bit Rate Mobile Landmark Search

AI Magazine

Coming with the ever growing computational power of mobile devices, mobile visual search have undergone an evolution in techniques and applications. A significant trend is low bit rate visual search, where compact visual descriptors are extracted directly over a mobile and delivered as queries rather than raw images to reduce the query transmission latency. In this article, we introduce our work on low bit rate mobile landmark search, in which a compact yet discriminative landmark image descriptor is extracted by using location context such as GPS, crowd-sourced hotspot WLAN, and cell tower locations. The compactness originates from the bag-of-words image representation, with an offline learning from geotagged photos from online photo sharing websites including Flickr and Panoramio. The learning process involves segmenting the landmark photo collection by discrete geographical regions using Gaussian mixture model, and then boosting a ranking sensitive vocabulary within each region, with an "entropy" based descriptor compactness feedback to refine both phases iteratively.


New application of machine learning and image analysis to help distinguish a rare subtype of kidney cancer: US researchers collaborate with scientist quarantined in China during COVID-19 outbreak

#artificialintelligence

Despite those obstacles, Indiana University School of Medicine faculty and Regenstrief Institute research scientists had their research published in Nature Communications on April 14, which is an even more significant feat considering one of the leading authors has been quarantined in Wuhan, China for the last two months of their work. The team consists of Affiliated Scientist Jie Zhang, PhD, Regenstrief Institute Research Scientist Kun Huang, PhD, both Indiana University School of Medicine faculty members, Jun Cheng, PhD, of Shenzhen University and colleagues including Liang Cheng, M.D. of IU School of Medicine. The study was led by Dr. Zhang, an assistant professor of medical and molecular genetics at IU School of Medicine. The work focuses on the application of machine learning and image analysis to help researchers distinguish a rare subtype of kidney cancer (translocational renal cell carcinoma, or tRCC) from other subtypes by examining the features of cells and tissues on a microscopic level. Dr. Zhang said the structural similarities have caused a high rate of misdiagnosis.


DSC Podcast Series: Machine Learning on the Edge

#artificialintelligence

Please Join this latest Data Science Central podcast to learn how you can develop machine learning models from an analytics base table and move that model to the edge for the purposes of scoring – or even training. Doing this will help you address latency reductions, security risks, and the potential for the corruption of your real time data. By improving models on the edge, your work will drive more data-driven business decisions faster.


xinshuoweng/AB3DMOT

#artificialintelligence

This repository contains the official python implementation for "A Baseline for 3D Multi-Object Tracking". However, recent works for 3D MOT tend to focus more on developing accurate systems giving less regard to computational cost and system complexity. In contrast, this work proposes a simple yet accurate real-time baseline 3D MOT system. We use an off-the-shelf 3D object detector to obtain oriented 3D bounding boxes from the LiDAR point cloud. Then, a combination of 3D Kalman filter and Hungarian algorithm is used for state estimation and data association.


Accelerating Medical Image Segmentation with NVIDIA Tensor Cores and TensorFlow 2 NVIDIA Developer Blog

#artificialintelligence

Medical image segmentation is a hot topic in the deep learning community. Proof of that is the number of challenges, competitions, and research projects being conducted in this area, which only rises year over year. Among all the different approaches to this problem, U-Net has become the backbone of many of the top-performing solutions for both 2D and 3D segmentation tasks. This is due to its simplicity, versatility, and effectiveness. When practitioners are confronted with a new segmentation task, the first step commonly is to use an existent implementation of U-Net as a backbone.


Sony announces camera sensor chips with built-in AI processing

#artificialintelligence

Sony has shown off what it's calling "the world's first image sensors to be equipped with AI processing functionality." These new sensors handle AI image analysis on board, so only the necessary data can be sent for further cloud processing. Artificial Intelligence is a natural pair with digital video cameras. They take in monstrous amounts of data, the vast majority of which is of no interest to anybody, particularly when you're talking about things like security cameras. As automation continues to escalate, we're going to need AI to keep an eye on more and more camera feeds.


IbrahimSobh/imageclassification

#artificialintelligence

In Transfer learning, we would like to leverage the knowledge learned by a source task to help learning another target task. For example, a well-trained, rich image classification network could be leveraged for another image target related task. Another example, the knowledge learned by a network trained on simulated environment can be transferred to a network for the real environment. A well known example for transfer learning is to load the already trained large scale classification VGG network that is able to classify images into one of 1000 classes, and use it for another task such as classification of special medical images. Image search engines: Generally speaking, search engine usually takes a query and returns results.


How should specialist physicians prepare for the AI revolution?

#artificialintelligence

Leaders in AI welcomed the report but called for greater collaboration between health organizations and researchers to advance the field. According to Hamid Tizhoosh, an engineer who leads the Laboratory for Knowledge Inference in Medical Image Analysis at the University of Waterloo, access to clinical data remains a challenge for developers. The task force recommendations are "all necessary, but none of them will help the advancement of AI in health care as much as the availability of data," Tizhoosh says. "What we need to see is a large number of large-scale initiatives and collaborations between hospitals, companies, and academic units to create the clinical data, look at the actual needs of doctors in hospitals, and then use AI to get things done."