Dreaming of Atmospheres

Waldmann, I. P.

doi:10.3847/0004-637x/820/2/107

Cited by 43 publications

(32 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While training, dropout helps prevent the network from over-fitting by randomly zeroing a neuron's output (Srivastava et al 2014). The use of dropout is to first-order equivalent to an L2 regularizer because it adaptively distorts the neuron data to control over fitting (Wager et al 2013).…”

Section: Neural Network Architecturementioning

confidence: 99%

Searching for exoplanets using artificial intelligence

Pearson

Palafox

Griffith

2017

Monthly Notices of the Royal Astronomical Society

141

101

View full text Add to dashboard Cite

In the last decade, over a million stars were monitored to detect transiting planets. Manual interpretation of potential exoplanet candidates is labor intensive and subject to human error, the results of which are difficult to quantify. Here we present a new method of detecting exoplanet candidates in large planetary search projects which, unlike current methods uses a neural network. Neural networks, also called "deep learning" or "deep nets" are designed to give a computer perception into a specific problem by training it to recognize patterns. Unlike past transit detection algorithms deep nets learn to recognize planet features instead of relying on hand-coded metrics that humans perceive as the most representative. Our convolutional neural network is capable of detecting Earth-like exoplanets in noisy time-series data with a greater accuracy than a least-squares method. Deep nets are highly generalizable allowing data to be evaluated from different time series after interpolation without compromising performance. As validated by our deep net analysis of Kepler light curves, we detect periodic transits consistent with the true period without any model fitting. Our study indicates that machine learning will facilitate the characterization of exoplanets in future analysis of large astronomy data sets.

show abstract

Section: Neural Network Architecturementioning

confidence: 99%

Searching for exoplanets using artificial intelligence

Pearson

Palafox

Griffith

2017

Monthly Notices of the Royal Astronomical Society

141

101

View full text Add to dashboard Cite

show abstract

“…There are several other machine learning methods that can be used to perform atmospheric retrieval (Waldmann 2016;Zingales & Waldmann 2018;Cobb et al 2019), each with their own advantages. We tested the same CCF-sequence retrieval as before, but now using a standard neural network and a standard Bayesian neural network (BNN) (Gal 2016).…”

Section: Comparison To Other Machine-learning Techniquesmentioning

confidence: 99%

Interpreting High-resolution Spectroscopy of Exoplanets using Cross-correlations and Supervised Machine Learning

Fisher

Hoeijmakers

Kitzmann

et al. 2020

View full text Add to dashboard Cite

We present a new method for performing atmospheric retrieval on ground-based, high-resolution data of exoplanets. Our method combines cross-correlation functions with a random forest, a supervised machine learning technique, to overcome challenges associated with high-resolution data. A series of cross-correlation functions are concatenated to give a "CCF-sequence" for each model atmosphere, which reduces the dimensionality by a factor of ∼ 100. The random forest, trained on our grid of ∼ 65, 000 models, provides a likelihood-free method of retrieval. The pre-computed grid spans 31 values of both temperature and metallicity, and incorporates a realistic noise model. We apply our method to HARPS-N observations of the ultra-hot Jupiter KELT-9b, and obtain a metallicity consistent with solar (log M = −0.2 ± 0.2). Our retrieved transit chord temperature (T = 6000 +0 −200 K) is unreliable as the ion cross-correlations lie outside of the training set, which we interpret as being indicative of missing physics in our atmospheric model. We compare our method to traditional nestedsampling, as well as other machine learning techniques, such as Bayesian neural networks. We demonstrate that the likelihood-free aspect of the random forest makes it more robust than nested-sampling to different error distributions, and that the Bayesian neural network we tested is unable to reproduce complex posteriors. We also address the claim in Cobb et al. (2019) that our random forest retrieval technique can be over-confident but incorrect. We show that this is an artefact of the training set, rather than the machine learning method, and that the posteriors agree with those obtained using nested-sampling.

show abstract

“…Combining this method with traditional transit spectroscopy provides a new avenue for high-fidelity atmospheric retrievals. There have also been efforts to try machine learning techniques such as artificial neural networks for retrievals (Waldmann 2016) but their efficacy on real datasets and benefits over state-of-the-art Bayesian inference methods remains to be seen.…”

Section: New Trends and Future Prospectsmentioning

confidence: 99%

Atmospheric Retrieval of Exoplanets

Madhusudhan

2018

Handbook of Exoplanets

View full text Add to dashboard Cite

Exoplanetary atmospheric retrieval refers to the inference of atmospheric properties of an exoplanet given an observed spectrum. The atmospheric properties include the chemical compositions, temperature profiles, clouds/hazes, and energy circulation. These properties, in turn, can provide key insights into the atmospheric physicochemical processes of exoplanets as well as their formation mechanisms. Major advancements in atmospheric retrieval have been made in the last decade, thanks to a combination of state-of-the-art spectroscopic observations and advanced atmospheric modeling and statistical inference methods. These developments have already resulted in key constraints on the atmospheric H 2 O abundances, temperature profiles, and other properties for several exoplanets. Upcoming facilities such as the JWST will further advance this area. The present chapter is a pedagogical review of this exciting frontier of exoplanetary science. The principles of atmospheric retrievals of exoplanets are discussed in detail, including parametric models and statistical inference methods, along with a review of key results in the field. Some of the main challenges in retrievals with current observations are discussed along with new directions and the future landscape.

show abstract

Dreaming of Atmospheres

Cited by 43 publications

References 48 publications

Searching for exoplanets using artificial intelligence

Searching for exoplanets using artificial intelligence

Interpreting High-resolution Spectroscopy of Exoplanets using Cross-correlations and Supervised Machine Learning

Atmospheric Retrieval of Exoplanets

Contact Info

Product

Resources

About