Welcome back, AI prodigies!

In today’s sunday special:

  • 📜 The Prelude

  • 🔭 A Brief History of Planet Hunting

  • 🛰️ How AI Helps Discover Hidden Planets

  • 🚀 Why Should We Care?

  • 🔑 Key Takeaway

Read time: 7 minutes

🩺 PULSE CHECK

Should we investigate whether life exists beyond Earth?

Vote Below To View Live Results

Login or Subscribe to participate

🎓 Key Terms

  • Deep Learning (DL): Mimics the structure and function of the human brain by processing data through multiple layers of artificial neurons.

  • Convolutional Neural Networks (CNNs): A network of specialized filters that detect complex visual patterns like edges and corners across images and videos.

📜 THE PRELUDE

The cold air bites at your cheeks, every breath visible in fleeting puffs, as the chilly night sky engulfs you in a boundless blanket of black. At first, each star seems pinned in place like distant jewels. But then, you begin to notice subtle shifts: a faint glimmer here, a fleeting drift there. It’s as if the cosmos itself were twirling in a galactic dance.

An exoplanet, or any planet beyond our solar system, is often too dim, too distant, or too docked to its host star to observe directly. Nevertheless, NASA confirmed the existence of 6,160 exoplanets across the universe. So, how exactly did NASA identify them?

In the scientific study of celestial objects, a transit occurs when a planet passes between a star and its observer, slightly dimming the star’s bright light. After drifting in deep space for nearly nine years, the Kepler / K2,” a space telescope assigned to search a portion of the Milky Way for Earth-sized planets, identified that our galaxy contains billions of hidden exoplanets, meaning the night sky is filled with more planets than stars!

That said, NASA quickly realized that stars wiggle, flicker, and twinkle, which mimic the tiny dips in brightness that occur during a transit, tricking space telescopes into incorrectly identifying stellar activity as an exoplanet. To address this, they developed a DL Model called ExoMiner,” which sorts real planets from imposter planets. So, how exactly does it accomplish this?

🔭 A BRIEF HISTORY OF PLANET HUNTING

⦿ 1️⃣ the Earth, the Sun, and the Center?

In 150 A.D., Roman astronomer Claudius Ptolemy proposed the Geocentric Model,” which placed a stationary Earth at the center of the universe, with all celestial bodies rotating around it. This worldview shaped not only astronomy but also theology and philosophy. The Catholic Church promptly endorsed this concept, with famous Italian priest Thomas Aquinas affirming the Earth-centered universe as divine order, aligning it with religious doctrine.

But in 1543, Polish polymath Nicolaus Copernicus proposed the Heliocentric Model,” where the Sun occupied the center of the universe, with the Earth orbiting around it in circular paths. Though initially met with skepticism, it gained strength through German mathematician Johannes Kepler, who described planetary orbits as ellipses, meaning the path a planet took around the Sun wasn’t a perfect circle but an elongated oval. Coupled with this, Italian astronomer Galileo Galilei’s refracting telescope observations showed moons orbiting Jupiter, proving that not everything orbits the Earth. The Catholic Church resisted fiercely, even prosecuting the Italian astronomer and forcing him to recant his planetary findings.

But in 1687, English polymath Sir Isaac Newton published PM,” where his three laws of motion and law of universal gravitation mathematically explained not only how objects move on Earth but also how the Moon orbits the Earth and how the Earth orbits the Sun. These foundational principles of physics forced humanity to accept that the Earth didn’t sit upon the universe’s throne.

⦿ 2️⃣ One Wobble at a Time?

For most of modern astronomy, exoplanets were merely figments of science fiction. But in 1995, Swiss astrophysicists Michel Mayor and Didier Queloz discovered 51 Pegasi b,” the first exoplanet orbiting a Sun-like star. For context, it’s a gas giant with an extremely small orbital cycle, completing a full year in just 4.2 days.

This scientific finding opened up an entirely new field of planetary discovery, proving not only that planets existed outside our solar system but also that planetary formations don’t always follow predictable orbital patterns. In simple terms, it established that planets were common, not rare.

Our solar system hosts two main types of planets: rocky bodies and gas giants. It was scientifically established that gas giants orbited farther from the Sun because they needed to form beyond the “frost line,” where it’s cold enough for volatile compounds to condense into solid ice grains and grow massive enough to attract gassy atmospheres.

51 Pegasi b completely upended this established scientific rule. It was the first “hot Jupiter” discovered, proving that while gas giants likely start far away where ice is abundant, they can undergo planetary migration, spiraling inward toward Sun-like stars after they form.

⦿ 3️⃣ From Wobbles to Shadows?

In 1999, American astronomer and renowned exoplanet explorer David Charbonneau pioneered the Transit Method”: measuring repeating dips in a host star’s brightness as an exoplanet passes in front of it. The depth of the dip tells us the exoplanet’s physical size. The time between each dip tells us the exoplanet’s orbital cycle.

He leveraged a small, ground-based telescope to watch over the host star HD 209458,” already known to host HD 209458 b,” a Jupiter-sized exoplanet. He repeatedly spotted the Jupiter-sized exoplanet transiting in front of the host star, causing consistent 1.5% dips in brightness every 3.5 days.

🛰️ HOW AI HELPS DISCOVER PLANETS

⦿ 4️⃣ Too Many Signals, Too Many Imposters?

In 2018, NASA launched TESS,” a surveying satellite designed to discover exoplanets within a celestial region 400x larger than Kepler / K2.” Together, these combined observational space missions collected more than 100,000 transit signals. In 2021, NASA’s Ames Research Center in California’s Silicon Valley developed ExoMiner,” which validates exoplanet candidates recorded across these observational space missions.

⦿ 5️⃣ How Exactly Does It Work?
  1. 🔴 Step #1: The Starlight Dips

    • TESS and Kepler / K2 continuously measure and monitor transit signals, with each transit signal producing a light curve that contains consistent dips, potentially indicating a passing exoplanet.

    • If any of these consistent dips meet a statistical threshold, known as the S/N Ratio,” which quantifies the strength of a potential exoplanet’s signals against the background noise of a host star, such as wiggles, flickers, and twinkles, they’re classified as a TCE.”

  2. 🟠 Step #2: The Objects of Interest

    • Not all TCEs are meaningful, so exoplanet scientists must further filter them by conducting basic biosignature false positives, such as whether the consistent dips repeat at stable intervals, matching expected transit shapes and signals.

    • TCEs that pass these basic biosignature false positives are then promoted to KOIs,” meaning the host star dims slightly at regular, repeatable, and recurring intervals. At this stage, the object of interest is labeled as a well-vetted, transit-like event.

  3. 🟡 Step #3: The Transit Training Phase

    • ExoMiner is trained on high-quality datasets of correctly identified inputs (e.g., transit signals) with their corresponding outputs (i.e., confirmed exoplanets). These high-quality training datasets often include:

      1. 📍 Time-Series Data: The brightness of a host star measured over time, represented as a numeric array like: “{1.000, 0.9998, 0.9996, 1.000....}.”

      2. 📌 Phase-Folded Views: The stacking of a host star’s brightness measurements, represented as a shorter numeric array that illustrates one average orbit or a single transit signal.

  4. 🟢 Step #4: The Layers of Analysis Phase

    • Once trained, ExoMiner passes the KOIs through a CNN, which scans the light curves through hundreds of specialized filters or “layers.” The earlier layers identify simple visual patterns like:

      1. 📐 Sudden Dips: Abrupt reductions in brightness.

      2. 📏 Repeated Troughs: Periodic dimming fluctuations.

  5. 🔵 Step #5: The Probability Score Phase

    • Once the CNN analyzes the light curves and extracts complex visual patterns indicative of an exoplanet, ExoMiner combines these findings with other raw input data, including transit depth and orbital period, to generate an initial probability score on a scale from 0 to 1, with 1 indicating complete confidence in the existence of an exoplanet.

🚀 WHY SHOULD WE CARE?

⦿ 6️⃣ “Are We Alone?” Becomes Science.

The nearest known exoplanets are located approximately 4.25 light-years away in Alpha Centauri: an exotic 3-star system that might host the first exoplanet with extraterrestrial life. For context, with current interstellar spacecraft technology, it would take roughly 165,000 years to reach it.

One of humanity’s most mystifying questions is, “Are we alone in the universe?” The latest galactic discoveries by exoplanet scientists reveal that planets beyond our solar system are abundant, and some may even harbor life.

If life is unique to Earth, then our planet’s biosphere is priceless. It’s either a statistical anomaly or a religious miracle. Either way, we have no choice but to treat Earth as both our first and our last habitable home.

If life is common outside our solar system, then it’s a scientifically predictable outcome rather than a statistical anomaly or a religious miracle. If biological rules repeat across the cosmos, the stories we tell ourselves will be rewritten.

🔑 KEY TAKEAWAY

Just 30 years ago, exoplanets were the stuff of science fiction. Today, aided by AI, we’ve confirmed the existence of exactly 6,160 exoplanets across 4,584 planetary systems. The study of exoplanets orbiting host stars outside our solar system can teach us how celestial bodies come into being and, ultimately, whether we’re alone in the universe.

📒 FINAL NOTE

FEEDBACK

How would you rate today’s email?

We read every reply you leave after rating!

Login or Subscribe to participate

❤️ Today’s Featured Reply

“The most value-packed AI newsletter I’ve read by far. I seriously can’t believe how young you guys are. Bravo! 👏

-Alfred (1️⃣ 👍 Nailed it!)
REFER & EARN

🎉 Your Friends Learn, You Earn!

{{rp_personalized_text}}

Share your unique referral link: {{rp_refer_url}}