ANC Workshop - Asa Cooper Stickland and Sigrid Hellan Speaker: Asa Cooper Stickland Title: Consistency of Neural Networks is Related to Loss Landscape Geometry Abstract: Recent work has shown the geometry of loss landscapes is important for generalization, in particular flatter minima tend to be better. I'll present an overview of some of this work and some of our own work showing a relationship between flat minima and using 'consistency losses' during training, particularly for multilingual NLP. Speaker: Sigrid Hellan Title: Bayesian Optimisation for Active Monitoring of Air Pollution Abstract: Air pollution is one of the leading causes of mortality globally, resulting in millions of deaths each year. Efficient monitoring is important to measure exposure and enforce legal limits. New low-cost sensors can be deployed in greater numbers and in more varied locations, motivating the problem of efficient automated placement. Previous work suggests Bayesian optimisation is an appropriate method, but only considered a satellite data set, with data aggregated over all altitudes. It is ground-level pollution, that humans breathe, which matters most. We improve on those results using hierarchical models and evaluate our models on urban pollution data in London to show that Bayesian optimisation can be successfully applied to the problem. Jan 25 2022 11.00 - 12.00 ANC Workshop - Asa Cooper Stickland and Sigrid Hellan Tuesday, 25th January 2022 online
ANC Workshop - Asa Cooper Stickland and Sigrid Hellan Speaker: Asa Cooper Stickland Title: Consistency of Neural Networks is Related to Loss Landscape Geometry Abstract: Recent work has shown the geometry of loss landscapes is important for generalization, in particular flatter minima tend to be better. I'll present an overview of some of this work and some of our own work showing a relationship between flat minima and using 'consistency losses' during training, particularly for multilingual NLP. Speaker: Sigrid Hellan Title: Bayesian Optimisation for Active Monitoring of Air Pollution Abstract: Air pollution is one of the leading causes of mortality globally, resulting in millions of deaths each year. Efficient monitoring is important to measure exposure and enforce legal limits. New low-cost sensors can be deployed in greater numbers and in more varied locations, motivating the problem of efficient automated placement. Previous work suggests Bayesian optimisation is an appropriate method, but only considered a satellite data set, with data aggregated over all altitudes. It is ground-level pollution, that humans breathe, which matters most. We improve on those results using hierarchical models and evaluate our models on urban pollution data in London to show that Bayesian optimisation can be successfully applied to the problem. Jan 25 2022 11.00 - 12.00 ANC Workshop - Asa Cooper Stickland and Sigrid Hellan Tuesday, 25th January 2022 online
Jan 25 2022 11.00 - 12.00 ANC Workshop - Asa Cooper Stickland and Sigrid Hellan Tuesday, 25th January 2022
