Open martiansideofthemoon opened 7 years ago
Most of the state-of-the-art use complex physics models. These (http://www.sciencedirect.com/science/article/pii/S0893608006000050) guys have used neural networks for weather prediction (sea level pressure, temperature, precipitation ... ) but have used some non-so-obvious parameters like CO2 and N2O levels. Here (http://download.springer.com/static/pdf/82/art%253A10.1007%252FBF00115009.pdf?originUrl=http%3A%2F%2Flink.springer.com%2Farticle%2F10.1007%2FBF00115009&token2=exp=1481116658~acl=%2Fstatic%2Fpdf%2F82%2Fart%25253A10.1007%25252FBF00115009.pdf%3ForiginUrl%3Dhttp%253A%252F%252Flink.springer.com%252Farticle%252F10.1007%252FBF00115009*~hmac=2688a45233f6a358c581502ed11a6edb7ffebc89dd05a2c4e80e42b9f4084b0f) is an interesting paper that has formulated the problem as an reinforcement learning problem and has used temporal difference learning.
Most of the state-of-the-art use complex physics models. These (http://www.sciencedirect.com/science/article/pii/S0893608006000050) guys have used neural networks for weather prediction (sea level pressure, temperature, precipitation ... ) but have used some non-so-obvious parameters like CO2 and N2O levels. Here (http://download.springer.com/static/pdf/82/art%253A10.1007%252FBF00115009.pdf?originUrl=http%3A%2F%2Flink.springer.com%2Farticle%2F10.1007%2FBF00115009&token2=exp=1481116658~acl=%2Fstatic%2Fpdf%2F82%2Fart%25253A10.1007%25252FBF00115009.pdf%3ForiginUrl%3Dhttp%253A%252F%252Flink.springer.com%252Farticle%252F10.1007%252FBF00115009*~hmac=2688a45233f6a358c581502ed11a6edb7ffebc89dd05a2c4e80e42b9f4084b0f) is an interesting paper that has formulated the problem as an reinforcement learning problem and has used temporal difference learning.