[Popular STEM] Curating the Internet: Science and technology digest for October 17, 2020

IEEE Spectrum's weekly selection of awesome robot videos; Extortionists are masquerading as international hacking groups; Machine learning and big-data are making earthquake prediction possible; A new cancer treatment is under study that works by acting like a virus; and An international team of doctors is criticizing Google for failing to back its claims on AI diagnosing breast cancer

Administrivia

Experimental: Please leave a relevant and substantive reply to this post in order to be considered for a gratuity from @penny4thoughts after the post pays out.

Links and micro-summaries from my 1000+ daily headlines. I filter them so you don't have to.

First posted on the Steem blockchain: SteemIt.

1. Video Friday: Agility Robotics Raises $20 Million to Accelerate Robot Production - IEEE Spectrum's weekly selection of awesome robot videos this week includes:
   • Digit,a biped from Agility Robotics that is designed to work around people and in tight spaces. It has path planning, perception, awareness of the world around us, and other aspects of "whole body behavior". The narrator suggests that legged robots like Digit will change the world as much as the automobile did.
   • DuAxel is a prototype rover from NASA that has the capability to travel long distances and rappel down to hard-to-reach places. This video shows the rover going through testing in California's Mojave Desert.
   • Tumbling Magnetic Microrobots from Purdue University. The video shows the microrobots navigating through a living colon.
   • The iCub biped robot is a child-sized robot that can walk and recover its balance, even after being pushed by a human.
   • and more
   
   
   Here is a stratospheric test flight of a Sunglider solar HAPS (high-altitude pseudo-satellite ) from AeroVironment.
   
   


2. Fancy Bear imposters are on a hacking extortion spree - Yesterday, here on Steem, @justyy reported an e-mail extortion attempt by someone using the name "Fancy Bear" who was threatening a DDoS attack if an extortion payment is not received. It turns out that he is not alone. According to the article, this group sometimes impersonates the Russian Fancy Bear hacking group and other times they impersonate a North Korean hacking group known as the Lazarus Group. In general, it seems that they claim to be Fancy Bear when targeting manufacturing or tech targets and the Lazarus Group when attacking financial targets. The article quotes the FBI as saying, "Most institutions that reached the six-day mark did not report any additional activity or the activity was successfully mitigated. However, several prominent institutions did report follow-on activity that impacted operations."


3. AI Shakes Up the Seismology World - Until recently, seismologist had basically given up on the idea of predicting earthquakes in the foreseeable future. There was too much data to process effectively, and the depth requirement for underground censors was prohibitively deep. The era of big-data and machine learning is changing this, however. The advent of big-data now makes it possible to process information from a vast number of inexpensive censors, including even accelerometers on people's smart phones. Similarly, the rise of machine learning makes it possible to separate the signal from the noise. In particular, researchers are now gaining insights into when a swarm of small shakes may or may not lead to a larger quake, and they are even able to distinguish between local and remote geologic activity and pick out false-positive ground motion caused by things like trucks, garage doors, and aircraft. As a result of these advances, researchers can now provide advance notice of 1-10 seconds for people to anticipate a quake and find a safe location to ride it out.


4. Drug tricks cancer cells by impersonating a virus - A new drug called, BO-112, is currently in human trials for treating cancer. The drug is designed to mimic a virus by simulating the structure of a double-stranded RNA molecule. This is done in order to trigger the body's immune system and enlist its capabilities to fight off the cancer. When a cell gets infected by a virus, or in this case by BO-112, the cell produces antigens which are observable from outside the cell. This causes the body to surge its immune cells to the location in order to eradicate the invader. Cancer cells, however, often have evolved mechanisms to conceal their presence from the immune system. Thus, when the cancer cell gets "infected" by BO-112 its like shining a spotlight on the cancer so that the body's immune system can fight it off. In addition to engaging the body's immune system, the treatment can also make it easier to target the cancer cells with other medical interventions. As-of now, the drug has been tested in mice and in a few dozen humans. According to the article,

   The early results hint that BO-112 can make tough-to-treat tumors vulnerable to immunotherapy, but the team now needs to confirm that those results hold up in larger groups.

   Dr. Joshua Brody, director of the Lymphoma Immunotherapy Program at the Icahn School of Medicine at Mount Sinai, is quoted as saying,

   The exciting opportunity presented by these two studies, both in the lab and in patients, is that we have medicines that can improve antigen presentation and thereby make immunotherapies — which would otherwise fail — become effective in inducing cancer remissions



5. Top doctors slam Google for not backing up incredible claims of super-human cancer-spotting AI - In a January paper in Nature, Google claimed that its AI is better than human doctors at spotting breast cancer in mammogram images. Now a team of human doctors have responded, complaining that Google didn't include enough evidence to support the claim. In particular, Dr Benjamin Haibe-Kains, lead author and senior scientist at the Princess Margaret Cancer Centre said,

   Without the computer code and fitted model, it will be very difficult to build on their work.

   He and 22 coauthors also wrote,

   publication of insufficiently documented research does not meet the core requirements underlying scientific discovery.

   and

   Merely textual descriptions of deep-learning models can hide their high level of complexity. Nuances in the computer code may have marked effects on the training and evaluation of results, potentially leading to unintended consequences. Therefore, transparency in the form of the actual computer code used to train a model and arrive at its final set of parameters is essential for research reproducibility.

And to help make Steem the go-to place for timely information on diverse topics, I invite you to discuss any of these links in the comments and/or your own response post.

Beneficiaries

• Experimental - penny4thoughts - 36%
• Public Steem API - justyy - 5%
• Public Steem API - steemit - 5%
• SteemWorld.Org Tools for Steemizens - steemchiller - 5%

About this series

Follow on steem: remlaps-lite, remlaps