Breakthrough Listen

The measurement of magnetic fields generated by skeletal muscle activity, called magnetomyography (MMG), has seen renewed interest from the academic community in recent years. Although studies have demonstrated complex models of MMG and experiments classifying between different movements using MMG, there has yet to be time frequency analysis of MMG as well as concurrent recordings of MMG and its electrical counterpart, surface electromyography (sEMG). Here, we aim to better understand MMG in the context of sEMG by simultaneously recording both modalities during various muscle contraction tasks. We found that, similar to sEMG, MMG shows highly linearly correlated power to the degree of muscle contraction, has a unimodal distribution in spectral power, and can detect changes in muscle fatigue via changes in the spectral distribution. One main difference we found was that MMG typically has more high frequency content compared to sEMG, even when accounting for the filtering induced by the size of the sEMG electrodes. We additionally demonstrate empirically the decrease in MMG power due to distance from the arm and show MMG decreases slower than the inverse square law and can be measured up to 50 mm from the surface of the skin. Finally, we were able to capture MMG with non-OPM sensors showing that sensor technology has made great strides towards enabling MMG applications.

We present a novel method for anomaly detection in solar system object data in preparation for the Legacy Survey of Space and Time. We train a deep autoencoder for anomaly detection and use the learned latent space to search for other interesting objects. We demonstrate the efficacy of the autoencoder approach by finding interesting examples, such as interstellar objects, and show that by using the autoencoder, further examples of interesting classes can be found. We also investigate the limits of classic unsupervised approaches to anomaly detection through the generation of synthetic anomalies and evaluate the feasibility of using a supervised learning approach. Future work should consider expanding the feature space to increase the variety of anomalies that can be uncovered during the survey using an autoencoder.