Today we have a guest post from Dr Eriita Jones and Professor Mark McDonnell. Eriita is a Planetary and Space Scientist, Research Fellow at the School of IT and Mathematical Sciences, University of South Australia, and an ECR member of the National Committee for Space and Radio Science. Her primary research areas are (i) the remote detection and characterisation of subsurface water environments on Mars and Earth, and (ii) quantifying the habitability of other planetary bodies. She is particularly interested in new computational data analysis techniques and in assessing the benefits of machine learning for space science. Mark McDonnell leads the Computational Learning Systems Laboratory at University of South Australia. He has published over 100 research articles in the fields of machine learning, computational neuroscience, and statistical physics. Mark has worked extensively with industry partners to deliver applied machine learning solutions in areas such as precision agriculture, recycling, and sports analytics. His research interests lie at the intersection of machine learning and neurobiological learning.
Artificial intelligence may get some bad press, but there are of course many tasks with which AI can provide tremendous benefit to human beings. One of the tasks that AI can be utilised for is called ‘image segmentation’, which is the process of automatically dividing an image into objects or categories so that every pixel in the image receives an associated label (e.g. car, dog, tree). This is essentially what the Planet Four citizen scientists are doing when they manually outline the boundaries to fans and blotches in polar springtime imagery from Mars. Just like a human being, in order to learn a new skill a machine needs to be taught (or ‘trained’) in the task it is being asked to perform. For state-of-the-art automated image segmentation, this training requires large amounts of data in the form of images with the categories of interest clearly labelled. In 2018, researchers at the Computational Learning Systems Laboratory at the University of South Australia in Adelaide, Australia, realised that large amounts of labelled imagery was exactly what the citizen scientists on the Planet Four project were generating. That was the start of a collaboration with the Planet Four Science Team. We wondered – could we teach an algorithm to automatically detect fans and blotches in Martian imagery? How well could a machine learn these complex features? And could the algorithm provide information which would assist the scientists in their study of these Martian phenomena?
The machine learning algorithms used here are examples of deep Convolutional Neural Networks (CNN’s) which generally perform very strongly on image segmentation problems. The algorithms are fed thousands of labelled fan and blotch images produced by the Planet 4 citizen scientists. After lots of exposure to what fans and blotches look like at different locations, years, solar longitudes, and resolutions, the algorithms become able to generalize from their experiences and apply their learning to new situations – in this case, unlabelled images that they have never seen before. In order to assess how well the machine learning techniques are performing, the algorithms are given a test. They are asked to predict where the boundaries of the fans and blotches are in some labelled images – but the algorithms are not shown the labels and have never seen those images before. We can then compare the machine’s predictions with the ‘correct answers’ – the manual labels drawn by citizen scientists. We compare with another method as well– a more traditional and less complex image classifier that does not employ machine learning. The figures below shows the output on a subset of one HiRISE image.
We are busily working on validating the output of the machine learning algorithms on a large number of images, but we can already see ways in which they can be very useful. Although the algorithms might not always find every fan or blotch in an image, they are very good at deciding whether there is at least one feature present. In other words, they do a good job at sorting out the images which have a fan or blotch, from those that have no fans or blotches at all. This is a very useful way of streamlining the presentation of images to the Planet Four Zooniverse platform – for example, instead of having to click through ‘featureless’ images the Planet FourTeam in future may wish to make sure that every image that appears will have a fan or blotch in it for labelling. Additionally, by automatically predicting the presence of fans and blotches in new images the algorithms provide early information on feature number and density that can allow the Planet Four team to be more selective in which images have the highest priority for manual labelling.
Could machine learning one day put citizen scientists out of a job? We don’t think this is very likely. The algorithms may eventually learn to perform very well on new images if those images are similar enough to ones they have seen before. But if they are shown an image that is very different (e.g with unusual lighting conditions, strange background terrain, or uncharacteristic fans and blotches), it is likely that the machine won’t be quite as good at segmentation as a well-trained human eye. So don’t worry citizen scientists, AI is just here to lend a hand – thanks for all the fabulous data, and stay turned for an exciting update in a few months!
If you check out the Planet Four website now, you will now see that the site is on hiatus with the retirement of their older platform. A huge thank to you everyone who has contributed to Planet Four over these past six years and especially in the past few weeks. We completed the sets of HiRISE images we needed to complete before the site was shutdown. Mission accomplished. Thank you so much!
This isn’t goodbye, it’s a see you soon. We’re learning so much from the Planet Four classifications/assessments that we’ll be back with more seasonal fans and blotches to map. We’re working on a new version of Planet Four with the Zooniverse’s new project builder platform, but it will take us time to build and test the new version of the project. In the meantime we have lots of classifications to analyze from the original site. The science team is currently working on four papers (!!!) based on your classifications, and this work will continue even if the website is paused. In addition, Planet Four: Terrains and Planet Four: Ridges are on the Zooniverse’s newer platform and will continue to be active.
We were formally informed this week by the Zooniverse, that the platform that Planet Four is hosted on will be retired and shutdown very soon. On April 30th, the platform will be shutdown. This means that the site will be on hiatus as the science team continues to work towards building a new version of the project on the Zooniverse’s project builder platform. This is going to take several months to complete, but in the meantime we are also analyzing your classifications and the Season 2 and 3 fan and blotch catalog. The science team has papers in the works and have been exploring possibilities/synergies with machine learning as well.
Planet Four’s future is bright. April 30th won’t be the end of Planet Four. We’ll be back. For at least the time being Planet Four Talk will be accessible and you can login and continue to make posts. Some point in the future, Planet Four Talk will become read only. Rest assured, your comments, hastags, and collections are stored in the Planet Four Talk database, so that information is saved and accessible to the science team for further investigation. We’ll keep you updated here on the blog.
Before we pause Planet Four, we need your help! We still have data on the site left to be classified. We believe we have found a connection between regional dust storms on Mars and the number of seasonal fans and blotches visible in a given spring season. To check whether or not these storms are playing a significant role, new images are live on the Planet Four website. We are hoping we can get as many images classified as possible before the site goes on hiatus.
Time is running out, and we need your help to get as many of these images classified before the old Zooniverse platform is retired on April 30th. If everyone classified 10 images, we’d be over the finish line. If you can spare a minute or two, please review images at http://www.planetfour.org.
The science team is working on migrating Planet Four to the Zooniverse’s more modern project builder (or panoptes) platform. This is a slow process because things are different in how the newer Zooniverse platform displays images and also we want to take the lessons we’ve learned over the past 6 years and use it to make the web interface even better. We thought we’d share some screen shots from our work-in-progress prototype.
The next stage will be getting some images on the site and beta testing the changes we want to make and seeing how well these tweaks do compared to the current Planet Four website/classification interface. This might take a few months, but we’re working hard to have this ready before the end of the year.
In the meantime, we have new images on the original Planet Four website that we are hoping to get classified before the older Zooniverse platform that runs the current Planet Four site is officially retired. We’re trying to make the push in April to get these new images classified. If you can spare a few minutes to classify an image or two on the main Planet Four site, we’d appreciate it.
Today we have a post by Candy Hansen, principal investigator (PI) of Planet Four and Planet Four: Terrains. Candy also serves as the Deputy Principal Investigator for HiRISE (the camera providing the images of spiders, fans, and blotches seen on the original Planet Four project). Additionally she is a member of the science team for the Juno mission to Jupiter. She is responsible for the development and operation of JunoCam, an outreach camera that involves the public in planning images of Jupiter.
We have discovered something very interesting in the number and size of the fans that show up on the south polar seasonal cap every spring, that you are measuring. It turns out that in springs following both global and regional type A dust storms we see a lot more fans than normal for that time of year. This picture compares sub-images from 7 Martian years taken in “Manhattan” at solar longitude 195-197. The position of Mars in its orbit is the solar longitude (“Ls”), and southern spring begins at Ls 180 when the sun crosses the equator and heads south. Mars years 29, 30 and 33 have visibly more fans. There was a global dust storm in Mars Year (MY) 28 that started in early summer. Intense Type A storms, which are regional and centered at high southern latitudes, took place in MY29 and MY32. It looks like the spring after these storms have large numbers of seasonal fans.
Although the visual impression is powerful when these images are compared we can go beyond that now, thanks to the Planet Four fan catalog that your work has populated. We can quantify the differences. We used the MY29 an MY 30 catalog that we’ve published this year in our first paper, and also newly generated catalogs for Manhattan for MY 28, MY31, MY 32. Instead of just saying “there are a lot more fans” we can say “there are over twice as many fans” in MY29 and MY30 compared to MY28, 31 and 32. We do that by querying the catalog – an example is shown below. The plot below shows numbers of fans as a function of time in the spring and we can compare 5 years at Ls 195. I had the pleasure of presenting this (your!) work at the 2019 Lunar and Planetary Science conference last week in Houston, Texas.
To confirm that Type A storms are playing a significant role in the composition of the seasonal ice sheet that produces the carbon dioxide jets that bring up the dust and dirt that create the seasonal fans and blotches, we need to look at the number of seasonal fans and the area covered in MY33. We only have classifications for Seasons 1-5 of the HiRISE seasonal monitoring campaign (MY28-32). This brings me to my request: We would really like to have Planet Four measurements for MY33. We have uploaded the images, so it is ready for you to process. We would like to thank you in advance for your generosity with your time. Once those measurements are in we will be ready to write our next paper documenting these findings in a peer-reviewed scientific journal. As you know we have published one paper already and two more are in progress. This is a significant result, and we could not have done this without all of you.
Help classify the new images of Manhattan today at http://www.planetfour.org.
Happy Birthday Planet Four. This month marks 6 years of Planet Four. We couldn’t do any of this without the Planet Four volunteer community. Thank you for all of your help and contributions. We hope you’re celebrating with a slice of cake or a serving of Mars pie. The team is really excited for what’s to come next. We’re working hard on follow-up papers to the first fan and blotch catalog release. We’re also starting preparations to move the project to the Zooniverse’s newer Project Builder Platform. We’ll keep you posted on all of these efforts right here on the blog. Lots more to come in 2019!
Greetings from Knoxville, Tennessee. Earlier this morning, I presented our first catalog and early results from comparing the fan directions over two Mars years at the American Astronomical Society’s Division for Planetary Science meeting. Here’s my slides.
After a hiatus, Planet Four: Ridges is back! We’ve get the second batch of Meridiani ridges search images live on the site. We’re finding from the analysis of the previous search classifications for regular polygonal ridges, that Planet Four: Ridges volunteers can identify polygonal ridges smaller than contained in previous catalogs. We expect the project can do the same for Meridiani ridges. Dive in today at https;//ridges.planetfour.org and classify an image or two.
You might have noticed that in July we added a tutorial to Planet Four: Terrains. After some changes to the front-end part of the Zooniverse platform, the team decided to add the tutorial. You’ll find it on the tab next to ‘Task’. There should be some new examples to help guide you while classifying. If you were a fan of the original help button and Spotter’s guide, don’t worry those are still available as well.
We’re in the middle of going through a new suite of images on the site. Dive in and check out the new tutorial and classify a few images today at http://terrains.planetfour.org.
We’ve uploaded new images to Planet Four: Terrains this week. This dataset continues to fill in areal coverage to look for spiders outside of the south polar layered deposits and also examining the overall distribution of spiders and other features.
You’ll notice some changes to the look of the classification interface. Over the past several months, the Zooniverse development team has made updates and changes to the classification pages. This Zooniverse Talk thread is where you can share your thoughts and feedback on the new look.
Dive in today a http://terrains.planetfour.org