Other Initiatives 🎉

Information Retrieval

At the last meeting at MSR, Devon proposed to use IR to retrieve images of climate change instead of generating them on the go. This can be pursued either as a standalone endeavour, or as complementary to the generative approach : i.e. when the viewer’s house is too different from the ones that our model is trained on, we can retrieve a photo of a similar house affected by extreme weather events. This would entail that the IR system would have to communicate with our backend.

Climate Conversational Agent

Even though a picture is worth a thousand words, it would be interesting to add other modalities to the tool, for instance for people who want more information about climate change on a global level or on a community level. For people who are interested on the project but who have a background in NLP, we can maybe create a climate conversational agent who can initially answer some basic questions about climate change, and eventually make it more complex, translating the climate model output into language, e.g. “According to the NOAA climate model, [your city] in 2100 will be highly impacted by tornadoes and flooding, with coast lines receding by an average of 5m. Since your house is further from the coast, you will be less impacted by the flooding itself, but the impact of strong winds …”