Generative deep learning can produce artificial, natural-looking images and other data, which has many promising applications in research — and in art. But the wide availability of generative models poses a challenge for society, which needs tools and best practices to distinguish between real and synthetic data.
Volume 2 Issue 3, March 2020
News & Views
Recurrent networks can be trained using a generalization of backpropagation, called backpropagation through time, but a gap exists between the mathematics of this learning algorithm and biological plausibility. E-prop is a biologically inspired alternative that opens up possibilities for a new generation of online training algorithms for recurrent networks.
As artists are beginning to employ deep learning techniques to create new and interesting art, questions arise about how copyright and ownership apply to those works. This Perspective discusses how artists, programmers and users can ensure clarity about the ownership of their creations.
Predicting the properties of batteries, such as their state of charge and remaining lifetime, is crucial for improving battery manufacturing, usage and optimisation for energy storage. The authors discuss how machine learning methods and high-throughput experimentation provide a data-driven approach to this problem, and highlight challenges in building models which provide fast and accurate battery state predictions.
With the aid of deep learning, the space of chemical molecules, such as candidates for drugs, can be constrained to find new bioactive molecules. A new open source tool can generate libraries of novel molecules with user defined properties.
Integrating knowledge about the circuit-level organization of the brain into neuromorphic artificial systems is a challenging research problem. The authors present a neural algorithm for the learning of odourant signals and their robust identification under noise, based on the architecture of the mammalian olfactory bulb and implemented on the Intel Loihi neuromorphic system.