In this study, explainable machine learning techniques are applied to predict the toxicity of mussels in the Gulf of Trieste (Adriatic Sea) caused by harmful algal blooms. By analysing a newly created 28-year dataset containing records of toxic phytoplankton in mussel farming areas and toxin concentrations in mussels (Mytilus galloprovincialis), we train and evaluate the performance of ML models to accurately predict diarrhetic shellfish poisoning (DSP) events. The random forest model provided the best prediction of positive toxicity results based on the F1 score. Explainability methods such as permutation importance and SHAP identified key species (Dinophysis fortii and D. caudata) and environmental factors (salinity, river discharge and precipitation) as the best predictors of DSP outbreaks. These findings are important for improving early warning systems and supporting sustainable aquaculture practices.

We present Trieste, an open-source Python package for Bayesian optimization and active learning benefiting from the scalability and efficiency of TensorFlow. Our library enables the plug-and-play of popular TensorFlow-based models within sequential decision-making loops, e.g. Gaussian processes from GPflow or GPflux, or neural networks from Keras. This modular mindset is central to the package and extends to our acquisition functions and the internal dynamics of the decision-making loop, both of which can be tailored and extended by researchers or engineers when tackling custom use cases. Trieste is a research-friendly and production-ready toolkit backed by a comprehensive test suite, extensive documentation, and available at https://github.com/secondmind-labs/trieste.

Soulless computer algorithms are already churning out weather bulletins, sports reports, rap lyrics and even passable Chinese poetry. But it seems machines have now taken another step towards replacing human enterprise by generating their own reviews of serious academic journal papers that are able to impress even experienced academics. Using automatic text generation software, computer scientists at Italy's University of Trieste created a series of fake peer reviews of genuine journal papers and asked academics of different levels of seniority to say whether they agreed with their recommendations to accept for publication or not. In a quarter of cases, academics said they agreed with the fake review's conclusions, even though they were entirely made up of computer-generated gobbledegook – or, rather, sentences picked at random from a selection of peer reviews taken from subjects as diverse as brain science, ecology and ornithology. "Sentences like'it would be good if you can also talk about the importance of establishing some good shared benchmarks' or'it would be useful to identify key assumptions in the modelling' are probably well suited to almost any review," explained Eric Medvet, assistant professor at Trieste's department of engineering and architecture, who conducted the experiment with colleagues at his university's Machine Learning Lab.