Recosys Remote Manual _hot_ · Fresh

| Paper | Key Idea | Why "Solid" | |-------|----------|--------------| | (Silveira et al., 2019 – ACM tutorial) | Offline vs online metrics, bias-vs-variance in A/B tests. | Must-read before any remote team runs experiments. |

| Paper | Key Idea | Why "Solid" | |-------|----------|--------------| | (Lee et al., 2019) | MAML (model-agnostic meta-learning) for few-shot personalization. | One of the few cold-start papers with public code and clear evaluation. | | "DropoutNet" (Volkovs et al., 2017) | Trains with item feature dropout to simulate cold items. | Simple, works with content features, easy to implement. | recosys remote manual

| Week | Topic | Paper(s) | |------|-------|-----------| | 1 | Basics | Koren 2009 (MF) + Hu 2008 (implicit) | | 2 | Ranking | Rendle 2009 (BPR) | | 3 | Deep learning | He 2017 (NCF) + Guo 2017 (DeepFM) | | 4 | Sequences | Kang 2018 (SASRec) | | 5 | Production | Covington 2016 (YouTube) | | 6 | Cold start | Volkovs 2017 (DropoutNet) | | Paper | Key Idea | Why "Solid"

Before you can control the device from a control room miles away, you must configure it locally. | One of the few cold-start papers with

Once programmed, the remote operates like a miniature air traffic controller for your living room:

To sync a new remote or a smart socket to your Recosys system:

: By pressing specific group buttons (like G-1 ), you activate a single switch while deactivating others, ensuring you don't accidentally dim the kitchen light when you meant to speed up the fan.