Articles in 2021

Predicting what comes next in a previously unseen time series of input data is a challenging task for machine learning. A novel unsupervised learning scheme termed predictive principal component analysis can extract the most informative components for predicting future inputs with low computational cost.

Identifying cancer driver genes from high-throughput genomic data is an important task to understand the molecular basis of cancer and to develop new treatments including precision medicine. To tackle this challenge, EMOGI, a new deep learning method based on graph convolutional networks is developed, which combines protein–protein interaction networks with multiomics datasets.

Rechargeable lithium-ion batteries play a crucial role in many modern-day applications, including portable electronics and electric vehicles, but they degrade over time. To ensure safe operation, a battery’s ‘state of health’ should be monitored in real time, and this machine learning pipeline, tested on a variety of charging conditions, can provide such an online estimation of battery state of health.

A generative approach called SliceGAN is demonstrated that can construct complex three-dimensional (3D) images from representative two-dimensional (2D) image examples. This is a promising approach in particular for studying microstructured materials where acquiring good-quality 3D data is challenging; 3D datasets can be created with SliceGAN, making use of high-quality 2D imaging techniques that are widely available.

Being able to precisely analyse behaviour is essential for the study of motor behaviour in health and disease, but is often time- and labour-intensive. Brattoli et al. develop an automatic approach based on deep learning for analysing motor behaviour and evaluate it on different species and diverse motor functions.

Identifying salient input features can be a challenge in neural networks. The authors developed a variable selection procedure with false discovery rate control that works on classification or regression problems, one or multiple output neurons, and deep or shallow neural networks.

The recent advances in computational chemistry rely on large data collections of chemical compounds and reactions. However, not all entries in these datasets are correct. Toniato and colleagues present here an automated approach to identify incorrect reactions, using the effect of catastrophic forgetting in neural networks.

The COVID-19 pandemic has highlighted key challenges for patient care and health provider safety. Adaptable robotic systems, with enhanced sensing, manipulation and autonomy capabilities could help address these challenges in future infectious disease outbreaks.

At the heart of many challenges in scientific research lie complex equations for which no analytical solutions exist. A new neural network model called DeepONet can learn to approximate nonlinear functions as well as operators.

It has been a little over a year since a worldwide COVID-19 pandemic was declared. Science has moved fast to fight the virus but preparations need to be underway for fighting future outbreaks.

Increased resolution of mass spectroscopy can provide better data for sequencing, but also increases the computational complexity of analysing the data. Qiao and colleagues present here a neural network-based method that processes sequencing data of any resolution while improving the accuracy of predicted sequences.

Neural networks are known as universal approximators of continuous functions, but they can also approximate any mathematical operator (mapping a function to another function), which is an important capability for complex systems such as robotics control. A new deep neural network called DeepONet can lean various mathematical operators with small generalization error.

The morphology of a robot determines how efficiently it can traverse different terrain. Nygaard and colleagues present here a robot that can adapt it’s morphology when it is detecting different terrain and learn which configuration is most effective.

At present, deep learning models in genomics are manually tuned through trial and error, which is time consuming and imposes a barrier for biomedical researchers not trained in machine learning. The authors develop an automated framework to design and apply convolutional neural networks for genomic sequences.

Many machine learning-based approaches have been developed for the prognosis and diagnosis of COVID-19 from medical images and this Analysis identifies over 2,200 relevant published papers and preprints in this area. After initial screening, 62 studies are analysed and the authors find they all have methodological flaws standing in the way of clinical utility. The authors have several recommendations to address these issues.

Neuromorphic computing could unlock low-power machine learning that can run on edge devices. A new algorithm that implements an artificial neuron emitting a sparse number of spikes could help realize this goal.

To truly understand the societal impact of AI, we need to look beyond the exclusive focus on quantitative methods, and focus on qualitative methods like ethnography, which shed light on the actors and institutions that wield power through the use of these technologies.

Imaging over a wide spectral profile allows for detailed chemical characterization, but processing this data is not trivial. A new neural network architecture can process hyperspectral information from a variety of imaging techniques effectively by using the popular U-net motive around another U-net processing spatial patterns.

Spiking neural networks could offer a low-energy consuming solution to deep learning applications on the edge and in mobile devices. Using temporal coding, where the timing of spikes carries extra information, a new method efficiently converts conventional artificial neural networks to spiking networks.

Cancers are complex diseases that are increasingly studied using a diverse set of omics data. At the same time, histological images show the interaction of cells, which is not visible with bulk omics methods. Binder and colleagues present a method to learn from both kinds of data, such that molecular markers can be associated with visible patterns in the tissue samples and be used for more accurate breast cancer diagnosis.