This is Harnham’s weekly news digest, the place to come for a quick breakdown of the week’s top news stories from the world of Data & Analytics.MIT News: A new state of the art for unsupervised computer visionComputer Vision models rely on labelling to identify objects, people and other characteristics. But labelling can be a lengthy process – producing just an hour of tagged and labelled data can take around 800 hours of human time.Scientists from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL), Microsoft, and Cornell University have been working to solve this problem by creating a new algorithm “STEGO” which learns “semantic segmentation”.Where a previous system might recognise the image of a dog playing in the park as just a dog, STEGO is able to assign a label to every pixel of the image and break it down into the main ingredients – a dog, sky, grass and its owner. To discover these objects without a human’s supervision. STEGO looks for similar objects that appear throughout a dataset. It then associates these objects together to construct a consistent view of the world across all the images it learns from.“The idea is that these types of algorithms can find consistent groupings in a largely automated fashion, so we don’t have to do that ourselves,” says PhD student Mark Hamilton.Read more here.KD Nuggets: 7 unique skills that set Data Scientists apart from other professions Data Science revolves around working with data, extracting insights, and communicating those insights, but to be a good Data Scientist honing your soft skills can be just as important.We have outlined the seven skills that separate data scientists from every other data job out there – and provided some easy way to improve them:Communication with your stakeholders – being able to tell a story behind the numbers and explaining technical problems with those with little or no, technical knowledge. Practice by explaining numeric things to your friends and family.Optimising for impact – the ability to identify and use the fastest, simplest method of answering a question is often overlooked. Simply ask yourself: is there an easier way to do this? And assume there almost always is.Patience – as a Data Scientist, you will often need to navigate businesspeople asking for technical things they don’t understand or are not possible and you may need to accept repeated misunderstandings of what you do. Practising answering questions calmly will really set you apart from your peers.Read more here.News Medical Life Sciences: Synthetic biologists interrogate single-celled survivalists to understand stress responseBacillus subtilis is a common soil bacterium with a strong reputation for handling stress. Biologists use this organism as a model to study spore formation – a life-or-death gamble that requires precision timing under the worst circumstances.Rice University synthetic biologists and bioengineers are preparing to interrogate B. subtilis’ stressed-out decision making. A new $1.3 million grant from the National Science Foundation will allow the team to use their optogenetic tools to decode B. subtilis’ genetic response to stress.“This research will reveal clues about how these pathways operate and give new insights into how bacteria survive stress.”“We also hope to uncover genetic design principles that help us better understand similar decision circuits in pathogenic bacteria” said Oleg Igoshin, professor of bioengineering and senior scientist at Rice’s Centre for Theoretical Biological Physics.Read more about their work here.PhysOrg: New machine learning model predicts how nanoparticles interact with proteins Antibiotic-resistant infections are on the rise, and as a continually morphing pandemic virus has highlighted, there is much work to be done to find ways to prevent these infections. Researchers at the University of Michigan have been developing a new Machine Learning model that predicts interactions between nanoparticles and proteins.The aim is to design medicines that can attack bacteria and viruses in ways that they choose, taking advantage of the “lock-and-key” mechanisms that dominate interactions between biological molecules. The new Machine Learning algorithm compares nanoparticles to proteins using three different ways to describe them. The two descriptors that concerned structure have become the most important for making predictions about whether a nanoparticle would be a lock-and-key match with a specific protein. The descriptions capture the protein’s complex surface and how it might reconfigure itself to enable lock-and-key fits.”There are many proteins outside and inside bacteria that we can target. We can use this model as a first screening to discover which nanoparticles will bind with which proteins,” said Emine Sumeyra Turali Emre, a postdoctoral researcher in chemical engineering.Read more here.We’ve loved seeing all the news from Data & Analytics in the past week, it’s a market full of exciting and dynamic opportunities. To learn more about our work in this space, get in touch with us at info@harnham.com.
