First Neural Network

It being 2017, there’s an insane amount of machine learning research in all sorts of directions and for all sorts of applications. Dealing with large amounts of data is exactly the purpose of machine learning, but ironically, sorting through all the noise of so much research makes the field difficult to enter without a guide. There have been several excellent guides written, and this is my process.

A few things are in agreement across the various help guides:

  • Anaconda is the best. Install it now and don’t ask questions. If some people disagree, that’s fine, but if you’re starting out, their quips with anaconda won’t apply to you. Installing the GUI version with Python 3 is a good starting point. All things are easily customizable later, so this doesn’t matter much.
  • PyTorch is the 2017 golden child, but being new, doesn’t have as much documentation yet. Starting out with something more documented will help; then you can go back to investigating PyTorch.
  • Keras is great for researchers, and you’ll need to choose TensorFlow or Theano as the back end. TensorFlow was designed to replace Theano and is a good starting choice if you’re undecided.
  • MachineLearningMastery.com is a fantastic source of tutorials. Several are linked below in a logical order.

 

Without further ado, here are a few tutorials that will let you get the basics down (all from Dr. Jason Brownlee at Machine Learning Mastery).

  • Set up python via anaconda
  • Machine Learning in Python (optional, or skip to Neural Network below)
    • In the #Load libraries module, replace this line:
      from pandas.tools.plotting import scatter_matrix
      with this one:
      from pandas.plotting import scatter_matrix
      …and that takes care of the FutureWarning error
    • I found that the SVM model was better than KNN (probably due to some package upgrades), so you can also add this code to your tutorial scripts to run that one:
      # Make predictions on validation dataset
      svm = SVC()
      svm.fit(X_train, Y_train)
      predictions = svm.predict(X_validation)
      print(accuracy_score(Y_validation, predictions))
      print(confusion_matrix(Y_validation, predictions))
      print(classification_report(Y_validation, predictions))
  • Set up Deep Learning
    • For OS X, install XCode if you get warnings.
    • This uses Keras + TensorFlow or Keras + Theano. I chose TensorFlow.
  • Neural Network tutorial
  • MNIST and CNNs
  • Using Keras on Amazon Web Services (AWS)

Astrophiz Podcast #36

There’s a first time for everything, and yesterday was the first time I was interviewed on a podcast!

Astrophiz usually covers astronomy and astrophysics, but they made an exception to interview a rocket scientist. For astronomy fans – don’t worry, there’s plenty more besides the interview. Check it out!


Mining the Moon for Rocket Fuel

I recently wrote this article for The Conversation with a few colleagues from the Caltech Space Challenge:

Gary Li, Ph.D. Candidate in Mechanical and Aerospace Engineering, University of California, Los Angeles
Danielle DeLattePh.D. Student in Aeronautics & Astronautics, University of Tokyo
Jerome GilleronPh.D. Candidate in Aerospace Engineering, Georgia Institute of Technology
Samuel WaldPh.D. Student in Aeronautics and Astronautics, Massachusetts Institute of Technology
Therese JonesPh.D. Candidate in Public Policy, Pardee RAND Graduate School

Special thanks to Sung Wha Kang, Rhode Island School of Design for creating the original images.

Between the Earth and the moon: An artist’s rendering of a refueling depot for deep-space exploration. Sung Wha Kang (RISD), CC BY-ND

Forty-five years have passed since humans last set foot on an extraterrestrial body. Now, the moon is back at the center of efforts not only to explore space, but to create a permanent, independent space-faring society.

Planning expeditions to Earth’s nearest celestial neighbor is no longer just a NASA effort, though the U.S. space agency has plans for a moon-orbiting space station that would serve as a staging ground for Mars missions in the early 2030s. The United Launch Alliance, a joint venture between Lockheed Martin and Boeing, is planning a lunar fueling station for spacecraft, capable of supporting 1,000 people living in space within 30 years.

Billionaires Elon Musk, Jeff Bezos and Robert Bigelow all have companies aiming to deliver people or goods to the moon. Several teams competing for a share of Google’s US$30 million cash prize are planning to launch rovers to the moon.

We and 27 other students from around the world recently participated in the 2017 Caltech Space Challenge, proposing designs of what a lunar launch and supply station for deep space missions might look like, and how it would work.

The raw materials for rocket fuel

Right now all space missions are based on, and launched from, Earth. But Earth’s gravitational pull is strong. To escape Earth’s gravity, a rocket has to be traveling 11 kilometers a second – 25,000 miles per hour!

Any rocket leaving Earth has to carry all the fuel it will ever use to get to its destination and, if needed, back again. That fuel is heavy – and getting it moving at such high speeds takes a lot of energy. If we could refuel in orbit, that launch energy could lift more people or cargo or scientific equipment into orbit. Then the spacecraft could refuel in space, where Earth’s gravity is less powerful.

The moon has one-sixth the gravity of Earth, which makes it an attractive alternative base. The moon also has ice, which we already know how to process into a hydrogen-oxygen propellant that we use in many modern rockets.

Roving Luna

NASA’s Lunar Reconnaissance Orbiter and Lunar Crater Observation and Sensing Satellite missions have already found substantial amounts of ice in permanently shadowed craters on the moon.

Those locations would be tricky to mine because they are colder and offer no sunlight to power roving vehicles. However, we could install big mirrors on the craters’ rims to illuminate solar panels in the permanently shadowed regions.

Mining operations on the moon, an artist’s rendering. Sung Wha Kang (RISD), CC BY-ND

Rovers from Google’s Lunar X Prize competition and NASA’s Lunar Resource Prospector, set to launch in 2020, would also contribute to finding good locations to mine ice.

Imagining a moon base

Depending on where the best ice reserves are, we might need to build several small robotic moon bases. Each one would mine ice, manufacture liquid propellant and transfer it to passing spacecraft. Our team developed plans to accomplish those tasks with three different types of rovers. Our plans also require a few small robotic shuttles to meet up with nearby deep-space mission vehicles in lunar orbit.

An artist’s rendering of lunar rover concepts. Sung Wha Kang (RISD), CC BY-ND

One rover, which we call the Prospector, would explore the moon and find ice-bearing locations. A second rover, the Constructor, would follow along behind, building a launch pad and packing down roadways to ease movements for the third rover type, the Miners, which actually collect the ice and deliver it to nearby storage tanks and an electrolysis processing plant that splits water into hydrogen and oxygen.

The Constructor would also build a landing pad where the small near-moon transport spacecraft we call Lunar Resupply Shuttles would arrive to collect fuel for delivery as newly launched spacecraft pass by the moon. The shuttles would burn moon-made fuel and would have advanced guidance and navigation systems to travel between lunar bases and their target spacecraft.

A gas station in space

An artist’s rendering of a fuel depot for refueling deep-space missions. Sung Wha Kang (RISD), CC BY-ND

When enough fuel is being produced, and the shuttle delivery system is tested and reliable, our plan calls for building a gas station in space. The shuttles would deliver ice directly to the orbiting fuel depot, where it would be processed into fuel and where rockets heading to Mars or elsewhere could dock to top up.

The depot would have large solar arrays powering an electrolysis module for melting the ice and then turning the water into fuel, and large fuel tanks to store what’s made. NASA is already working on most of the technology needed for a depot like this, including docking and fuel transfer. We anticipate a working depot could be ready in the early 2030s, just in time for the first human missions to Mars.

To be most useful and efficient, the depot should be located in a stable orbit relatively near both the Earth and the moon. The Earth-moon Lagrangian Point 1 (L1) is a point in space about 85 percent of the way from Earth to the moon, where the force of Earth’s gravity would exactly equal the force of the moon’s gravity pulling in the other direction. It’s the perfect pit stop for a spacecraft on its way to Mars or the outer planets.

Leaving Earth

Our team also found a fuel-efficient way to get spacecraft from Earth orbit to the depot at L1, requiring even less launch fuel and freeing up more lift energy for cargo items. First, the spacecraft would launch from Earth into Low Earth Orbit with an empty propellant tank.

An artist’s rendering of a solar electric propulsion tug above an asteroid. NASA

Then, the spacecraft and its cargo could be towed from Low Earth Orbit to the depot at L1 using a solar electric propulsion tug, a spacecraft largely propelled by solar-powered electric thrusters.

This would let us triple the payload delivery to Mars. At present, a human Mars mission is estimated to cost as much as US$100 billion, and will need hundreds of tons of cargo. Delivering more cargo from Earth to Mars with fewer rocket launches would save billions of dollars and years of time.

A base for space exploration

Building a gas station between Earth and the moon would also reduce costs for missions beyond Mars. NASA is looking for extraterrestrial life on the moons of Saturn and Jupiter. Future spacecraft could carry much more cargo if they could refuel in space – who knows what scientific discoveries sending large exploration vehicles to these moons could enable?

By helping us escape both Earth’s gravity and dependence on its resources, a lunar gas station could be the first small step toward the giant leap into making humanity an interplanetary civilization.

Editor’s Note: This story was updated to clarify the distinction between escape velocity and the velocity needed to achieve orbit.


Starting New Space Events in Tokyo

I write this, about two years into a journey that started bearing fruit two months ago. The best-organized events look effortless and operate like clockwork, but rarely is the scene behind the curtain so perfect. Back in late 2014, I had approached the existing ISU Space Cafe – DC lead, Angela Peura, about two things: (1) the event was begging for a website (and a Facebook page) and (2) I was moving to Tokyo and wanted to take the Space Cafe idea with me. She agreed to both and as I started preparing for the move to Tokyo, I assumed the Tokyo version of Space Cafe would be up and running within three months of my August 2015 arrival. Haha.

When I got to Tokyo, several elements of the landscape became clear.
(1) I had no concept of what events looked like in Tokyo
(2) I didn’t know which bars or venues would be friendly to our idea
(3) I didn’t know anyone in the space community who lived in Tokyo

Whoops.

The first year in Tokyo was filled with all the newness and confusion that comes with moving to the literal other side of the planet and attempting to decipher kanji. I started attending comedy and storytelling events put on by the expat community. Not a whole lot happened with Space Cafe, but I did manage to start SEDS UTokyo (another story…) and meet some cool people. Around my one year mark, two events started and popped on my radar: Nerd Nite and Perfect Liars Club. Things picked up steam as I approached the owner of a delightful British pub in Shimokitazawa about having our event there.

Venue done, now to find speakers…

As it turns out, the founders of Perfect Liars Club had been transplants from DC themselves, and (sympathetically) they introduced me to our first speaker, Elizabeth. The rest, as they say, is history and we picked up momentum once people started hearing about us.

From our first night, we had a good size crowd of a healthy mix of space enthusiasts and professionals. It turns out that there was a real desire in the Tokyo space community for an event like this, which is amazing and gratifying. We (now there’s a team!) even started a meetup.com group to help consolidate space-related events in Tokyo and are planning to expand from (slightly structured) Space Cafes to (even more informal) discussion happy hours in the coming months.

So to anyone who is interested in starting their own STEAM/outreach events – do it!

…but be patient. These things take way more time than you might think.


Uncommon Space Paths

As senior year of college draws to a close, many US students are considering one of two main paths: industry or academia. Usually the search is conducted exclusively within the US and few students look into the variety of international masters programs, and internships that exist in the rest of the world.

Of course, this isn’t an exhaustive list. What other programs do you know about or recommend?

 

 


Finding Your Tribe

Since beginning a graduate program in Tokyo, I have gotten countless questions and comments that are some flavor of “how are you adjusting to Japan?” and “I would never pack up and move halfway around the world.” The truth, as always, is complicated. Going to new places where you don’t know anyone is always a challenge, and there is always an adjustment period. Whether that is a week, month, year, or decade depends on the specific factors of the country, your personality, language barriers, and whether you have found fulfillment in the new location.

In Japan, I sometimes talk to prospective undergraduate students who want to study in the US, and the same question comes up each time: do you have any advice for adjusting to America? Having gone through my own adjustment over the last year and a half, I found this question initially hilarious but absolutely true. Any big culture shift is going to be a challenge.

The most important bit of advice I could impart is FIND YOUR TRIBE.

This could mean finding local people who speak your language. It could be people who share a passion or sport. It could be people who like to sit together while reading in coffee shops. Take your time, try a variety of events and settings. These things do not usually come together overnight. When nothing seems to be fitting, try starting your own event or group. You may find that there is a need for community that you can fill, and you may find that the tribe was there all along.

One lesson I learned over and over was that the information was out there, but I hadn’t found it yet. Want a group that does machine learning together? There’s a meetup.com group for that. Want to nerd out over beer? There’s a monthly event. Sometimes there are barriers in group dynamics – need for a promoter, need for a graphics designer, need for a photographer – so don’t be shy. You may be able to enable someone else to create something.

The new normal will be different and may require effort. It may take an hour each way to get there on very crowded trains, but when you find your tribe, you’ll know.


Art and Space

Art has played a huge role in how we think about space. Space heroes of the past dabbled as artists and created fantastic visions that we have spent the past few decades making into reality. These images from the 1970s may look familiar if you’ve seen Elysium.

All images courtesy of NASA Ames, no copyright.

 


Things Students Ask

Sometimes in outreach, engineers can feel like there’s a barrier to entry. They might be nervous about what kids might ask or whether they’ll look silly. So, for your information and enjoyment, here are some of my favorite questions I’ve ever been asked and answers that I may or may not have pieced together, but in retrospect would have been the best.

I’ve given a few tours for my group at NASA, and kids have asked some great questions. (Shout out to the students in Ecuador for what was definitely the best collection of questions I’ve ever heard!)

Do you believe in aliens? This is highly personal, but you can go in a few directions, and there’s always this song by Hank Green.

How do astronauts go to the bathroom in space?

Do you know what Area 51 is?

Why is the robot moving so slowly? To keep engineers safe

How many satellites are in space? About 1000 active satellites

 

And some other common questions that would be personal:

What’s it like to be a female in engineering?/Did anyone ever tell you you couldn’t be an engineer because you’re a girl?

What’s your favorite project you’ve ever done?

Where did you go to school?

What do you do?


User Interface Design

A few things I’ve learned about user interface design working on a flight project at NASA:

  • Assume your operator is looking at your Graphical User Interface (GUI) at 3am after having been awake since 6am the previous day.
  • Make stop buttons tiny/hidden to keep them from getting accidentally pressed.
  • Make ABORT buttons huge.
  • In your user guide, explain how each value on the GUI is calculated. Is it a raw value from telemetry or a processed one?
  • Make fonts larger than you think they should be.
  • Set each font individually in LabView or it will change when you move applications to a different computer.