Martin Gauch

DL & Earth Science
Google Research

[Yossi Matias @ymatias]

Advancing flood forecasting for those on the front lines: today we unveil enhanced Flood Hub with new features designed to empower experts in aid, government & research for better predictions & preparedness.
#FloodForecasting #AIforClimate #GoogleResearch
https://t.co/XKda9l8FTx

[Martin Gauch @martingauch]

I'll be giving a talk @EOTECDevNet next week, covering some recent(-ish) updates to Google's flood forecasting system, including new model, data, and coverage. Details ⬇️

[EOTEC DevNet @EOTECDevNet]

🌍 Join EOTEC DevNet’s January 2025 Floods Working Groups! 🌊

Learn about Google Flood Hub, an AI-powered flood forecasting tool, at our first meeting of the year.

🗓️ Dates: Jan 15-16, 2025

🔗 Learn more: https://t.co/j7FKiiTyVf

#Floods #CapacityBuilding #EOTECDevNet

[Martin Gauch @martingauch]

I'll be at #AGU24 as well, looking forward to seeing some neuralhydrology (or non-neural or non-hydrology) research!

[Google AI @GoogleAI]

Floods are among the most widespread natural disasters. Here we describe our new flood forecasting model for state-of-the-art 7-day lead time prediction, expanded forecasts on ungauged watersheds for expert use, and a new reanalysis & re-forecast dataset. https://t.co/mJZWxkpD7C

[Daniel Klotz @ido87]

Hey there 👋

Want to do a Phd with me? Apply: https://t.co/da5Wb0yagW

[Maximilian Beck @maxmbeck]

The #xLSTM is finally live! What an exciting day!
How far do we get in language modeling with the LSTM compared to State-of-the-Art LLMs?
I would say pretty, pretty far!
How? We extend the LSTM with Exponential Gating and parallelizable Matrix Memory!
https://t.co/Z2xaH0wfji

[Google AI @GoogleAI]

Large-scale global flood forecasting has been out of reach for a long time. In our Nature paper published today we show how breakthroughs in AI can close the gap & provide reliable flood predictions even in regions that previously lacked data. Learn more: https://t.co/A35HyTczZr

[HESS Journal @EGU_HESS]

Discussion open for comments: HESS Opinions: Never train an LSTM on a single basin https://t.co/VqlcBxaTNH

[Martin Gauch @martingauch]

Cool #NeurIPS2023 work on time series uncertainty prediction, led by @AndAuer
(including a small hydrology application - stay tuned for more!)

[Andreas Auer @AndAuer]

Interested in reliable prediction intervals for time series? 🚀

Excited to announce our latest work, which proposes HopCPT, was accepted at #Neurips2023 🎉. HopCPT tackles the challenges of time series data for Conformal Prediction with the help of Modern Hopfield Networks.
1/4

[Martin Gauch @martingauch]

I have to correct myself, one of them calls itself a llama (or LamaH), which are humpless. So only up to four humps. Still impressive.

[Martin Gauch @martingauch]

Three new camels joining the caravan in one day 🚀 Depending on the type of camel, that's up to six humps!

[Nans Addor @NansAddor]

A CAMELS dataset for hydrological Switzerland and a preprint are now available! https://t.co/51GfIOYPCe and https://t.co/OaOQCcJ7Kp. A thread.

[Fabian Paischer @PaischerFabian]

Excited to share our latest work on a semantic and interpretable memory module for RL! Complementary to recent developments in the realm of explainable AI, we focus on interpretability w.r.t. the memory of an agent.
1/n

[Martin Gauch @martingauch]

Here's a thread with the main results:
https://t.co/6KOV63m4br

[Martin Gauch @martingauch]

Remember when we asked you to rate hydrographs?

We just published the preprint with results!

"In Defense of Metrics: Metrics Sufficiently Encode Typical Human Preferences Regarding Hydrological Model Performance"
➡️ https://t.co/cHseC49geL

Main results in thread 🧵 1/7

[Martin Gauch @martingauch]

It's been a while since we first asked people to participate, but many thousand ratings later the "Rate my Hydrograph" paper is now published in WRR! 🎉

📄 Paper: https://t.co/I7ZpCCVE2x
📈 Code & data: https://t.co/Z9Px7P1TpM

Huge thanks to everyone who participated!

[Yossi Matias @ymatias]

Today, thanks to advances in AI-based global models, we're expanding our flood forecasting coverage to 80 countries for areas where 460M people live. We hope this will help governments, NGOs and people at risk to take timely action.
https://t.co/90EfiQTiPB

[Martin Gauch @martingauch]

More 🐫 joining the Caravan! 🎉

[Yossi Matias @ymatias]

We are collaborating with the Research ecosystem to advance AI and ML toward better Flood Forecasting and other hazard mitigation. One outcome of our collaboration workshops is the Caravan project - an open-source repository for global streamflow data

[CoLLAs 2026 @CoLLAs_Conf]

We are happy to announce the next 2 editions of CoLLAs! CoLLAs 2023 will happen once again in Montreal, Canada and CoLLAs 2024 will happen in Italy!

[Martin Gauch @martingauch]

Working on Deep Learning in Hydrology? Abstract submissions to our #EGU23 session are now open!