Thread by @nipun_batra

Happy to announce our @RealAAAI #aaai2022 paper titled "Accurate and Scalable Gaussian Processes for Fine-grained Air Quality Inference" with @patel_zeel_ @shivam15sahni, @PalakPurohit18 and @HarshP_ got accepted. Here is the 3 year old story on this work. (1/n)

@RealAAAI @patel_zeel_ @shivam15sahni @PalakPurohit18 @HarshP_ We started the work in 2018 with @deepaknaray12 a third year undergrad at @cse_iitgn. We dabbled around with various baselines for an year without much "external output" (2/n)

@RealAAAI @patel_zeel_ @shivam15sahni @PalakPurohit18 @HarshP_ @deepaknaray12 @cse_iitgn In the summer of 2019, @ApoorvAgnihotr2 also joined the team. While exploring other baselines, we started studying Gaussian processes and ended up submitting an article to @distillpub later that summer. (3/n)

@RealAAAI @patel_zeel_ @shivam15sahni @PalakPurohit18 @HarshP_ @deepaknaray12 @cse_iitgn @ApoorvAgnihotr2 @distillpub Fall 2019 onwards, @deepaknaray12 and @ApoorvAgnihotr2 worked very hard for us to submit an initial version to ACM IMWUT (Ubicomp). It got desk rejected within a month as out of scope. (4/n)

@RealAAAI @patel_zeel_ @shivam15sahni @PalakPurohit18 @HarshP_ @deepaknaray12 @cse_iitgn @ApoorvAgnihotr2 @distillpub The timely desk rejection helped us submit to KDD in Feb 2020. The paper was rejected with a lot of critical useful feedback. (5/n)

@RealAAAI @patel_zeel_ @shivam15sahni @PalakPurohit18 @HarshP_ @deepaknaray12 @cse_iitgn @ApoorvAgnihotr2 @distillpub Now, PhD student @patel_zeel_ also started working with @deepaknaray12 and @ApoorvAgnihotr2 We had done a lot of work on baselining air quality inference and created a toolkit for it. We submitted the toolkit work to ACM Sensys. It got rejected with a lot of suggestions. (6/n)

@RealAAAI @patel_zeel_ @shivam15sahni @PalakPurohit18 @HarshP_ @deepaknaray12 @cse_iitgn @ApoorvAgnihotr2 @distillpub In between due to COVID some workshops at venues like ICML, NeurIPS, became virtual. This allowed us to submit work we earlier would not have (travel, registration, etc. hard to justify for workshops when limited budget). We got good feedback from various such venues. (7/n)

Resubmitted the ACM Sensys rejected work at ACM/IEE E IPSN with improvements. Got rejected. Hard to get toolkits papers out?! (8/n)

In between all this, our @distillpub post got accepted: distill.pub/2020/bayesian-… (9/n)

Now, we are in 2021 Fall. @patel_zeel_ has been working hard. Three undergrads @shivam15sahni, @PalakPurohit18 and @HarshP_ also join the team. (10/n)

I ask the undergrads to create an interactive article (similar but smaller to Distill ones) for VISXAI workshop. They work hard, but, it gets rejected. They learn concepts while trying to create the interaction. (11/n)

In Sept. 2021, the team works hard and is able to submit to AAAI. We make several critical discoveries days before the submission. The team has been working hard over the past 2.5 years with tools and this preparation makes it easy to translate ideas to code! (12/n)

Post submission, we anticipate rebuttal questions and work on a (small) subset of those. Should have done more! (13/n)

The rebuttal period is 4 days and around Diwali! Our scores are mostly encouraging. Accept, Weak Accept, Weak Accept, Borderline Reject. (14/n)

The team works very hard for the rebuttal and is able to do all suggested improvement: more baselines, more analysis, more metrics, better explanation! It is an incredible performance. I proclaim - I have not seen a better rebuttal! Fingers crossed now! (15/n)

Today afternoon we get the notification of paper acceptance. We are very happy that the rebuttal helped. Paraphrasing the meta reviewer: In the rebuttal, the authors added experiments and evidence for each of these issues, which substantially addresses the critiques. (16/n)

In between all this, we were fortunate to get feedback: from Zeel's committee: @adagu and @uditbhatiaiitgn and various places we submitted to. (17/n)

Highlighted the journey, so that we remain grounded and those reading may feel encouraged! (N/N)