A recent paper published in National Science Review presents a multi-space alignment approach for cross-species and cross-modality electroencephalogram (EEG) based epileptic seizure detection. By employing deep learning techniques, including domain adaptation and knowledge distillation, it aligns cross-species and cross-modality EEG signals to enhance the detection capability beyond traditional within-species and within-modality models. Experiments on multiple scalp and intracranial EEG datasets of humans and canines demonstrated substantial improvements in the detection accuracy, achieving over 90% AUC scores for cross-species and cross-modality seizure detection with extremely limited labeled data from the target species/modality. This is the first study that demonstrates the effectiveness of integrating heterogeneous data from different species and modalities to improve EEG-based seizure detection performance.

The capability to switch “on” and “off” between infrared stealth states has been regarded as a future technology and is expected to play an important role in defense as well as many other fields. Towards this goal, Scientist in China used the non-volatile phase change material Ge₂Sb₂Te₅(GST) and proposed a switchable infrared stealth bilayer metafilm. By adjusting the states of GST, the infrared radiation characteristics of the target can be flexibly controlled between the infrared stealth and the non-stealth states. The technique will promote the research towards practical applications of intelligent infrared stealth technology.

Scientists developed a new way of investing in stocks using natural language processing. Using dynamic topic modeling, a variant of Latent Dirichlet Allocation, the new model uncovers hidden risk factors directly from company reports and translates them into tradable indices with minimal human oversight. Investors can now trade these risk factors directly and track industry trends using only the information contained in words.