Reviews & Analysis

Traditional sensing techniques apply computational analysis at the output of the sensor hardware to separate signal from noise. A new, more holistic and potentially more powerful approach proposed in this Perspective is designing intelligent sensor systems that ‘lock-in’ to optimal sensing of data, making use of machine leaning strategies.

Online targeted advertising fuelled by machine learning can lead to the isolation of individual consumers. This problem of ‘epistemic fragmentation’ cannot be tackled with current regulation strategies and a new, civic model of governance for advertising is needed.

Drug repurposing provides a way to identify effective treatments more quickly and economically. To speed up the search for antiviral treatment of COVID-19, a new platform provides a range of computational models to identify drugs with potential anti-COVID-19 effects.

Modern machine learning approaches, such as deep neural networks, generalize well despite interpolating noisy data, in contrast with textbook wisdom. Mitra describes the phenomenon of statistically consistent interpolation (SCI) to clarify why data interpolation succeeds, and discusses how SCI elucidates the differing approaches to modelling natural phenomena represented in modern machine learning, traditional physical theory and biological brains.

A challenge for multiscale simulations is how to link the macroscopic and microscopic length scales effectively. A new machine-learning-based sampling approach enables full exploration of macro configurations while retaining the precision of a microscale model.

Deep learning applied to genomics can learn patterns in biological sequences, but designing such models requires expertise and effort. Recent work demonstrates the efficiency of a neural network architecture search algorithm in optimizing genomic models.

State of the art neural network approaches enable massive multilingual translation. How close are we to universal translation between any spoken, written or signed language?

Hyperspectral imaging can reveal important information without the need for staining. To extract information from this extensive data, however, new methods are needed that can interpret the spatial and spectral patterns present in the images.

Medical artificial intelligence and machine learning technologies marketed directly to consumers are on the rise. The authors argue that the regulatory landscape for such technologies should operate differently when a system is designed for personal use than when it is designed for clinicians and doctors.

The dynamical properties of a nonlinear system can be learned from its time-series data, but is it possible to predict what happens when the system is tuned far away from its training values?

3D image reconstruction is important for the understanding of materials and their function in devices. A generative adversarial network architecture reconstructs 3D materials microstructures from 2D images.

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.

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.

Computational models that capture the nonlinear processing of the inner ear have been prohibitively slow to use for most machine-hearing systems. A convolutional neural network model replicates hallmark features of cochlear signal processing, potentially enabling real-time applications.

Many researchers have become interested in implementing artificial intelligence methods in applications with socially beneficial outcomes. To provide a way to study and benchmark such ‘AI for social good’ applications, Josh Cowls et al. use the United Nations’ Sustainable Development Goals to systematically analyse AI for social good applications.

The Conference on Neural Information Processing Systems (NeurIPS) introduced a new requirement in 2020 that submitting authors must include a statement on the broader impacts of their research. Prunkl and colleagues discuss challenges and benefits of this requirement and propose suggestions to address the challenges.

Chemical reactions can be grouped into classes, but determining what class a specific reaction belongs to is not trivial on a large-scale. A new study demonstrates data-driven automatic classification of chemical reactions with methods borrowed from natural language processing.

Evolutionary computation is inspired by biological evolution and exhibits characteristics familiar from biology such as openendedness, multi-objectivity and co-evolution. This Perspective highlights where major differences still exist, and where the field of evolutionary computation could attempt to approach features from biological evolution more closely, namely neutrality and random drift, complex genotype-to-phenotype mappings with rich environmental interactions and major organizational transitions.

The popularity of deep learning is leading to new areas in biomedical applications. Wang and colleagues summarize in this Review the recent development and future directions of deep neural networks for superior image quality in the tomographic imaging field.

DNN classifiers are vulnerable to small, specific perturbations in an input that seem benign to humans. To understand this phenomenon, Buckner argues that it may be necessary to treat the patterns that DNNs detect in these adversarial examples as artefacts, which may contain predictive information.