Luiz Felipe Vecchietti

Ph.D. KAIST. Researcher interested in Reinforcement Learning, Deep Learning, Data Science and Sports Analytics.

[Luiz Felipe Vecchietti @lfvecchietti]

nothing

[Luiz Felipe Vecchietti @lfvecchietti]

@LucianaBenotti @FortunatoMeire @_LXAI Thanks!

[Meire Fortunato @FortunatoMeire]

Come to say hi in the LatinX Social @_LXAI happening later on today (starting 3PM London time, 11am in Brazil 🇧🇷🇧🇷🇧🇷)

[Luiz Felipe Vecchietti @lfvecchietti]

@GenePark @ElTedao @VecchiettiLuis

[Aravind Srinivas @AravSrinivas]

New paper - CURL: Contrastive Unsupervised Representations for RL! We use the simplest form of contrastive learning (instance-based) as an auxiliary task in model-free RL. SoTA by *significant* margin on DMControl and Atari for data-efficiency. https://t.co/Xszek8cFFs

[Danny Lange @danny_lange]

Wow! @unity3d ML-Agents in a Sim2Real scenario. Absolutely amazing work 💥💥💥 If you want to try ML-Agents for yourself: https://t.co/GtVhgi7YYa ⚽️ #ReinforcementLearning #ArtificialIntelligence

[Luiz Felipe Vecchietti @lfvecchietti]

@TL_Metne Pra te confortar primo: @ElTedao essa semana conseguiu perder com o Daejeon, com o Cerezo e com o Botafogo. Cada um de uma forma diferente. Hahahahha

[Luiz Felipe Vecchietti @lfvecchietti]

@FortunatoMeire @Khipu_AI Imperdivel. 5am aqui na Coreia do Sul mas prestigiarei.

[Luiz Felipe Vecchietti @lfvecchietti]

Inspiring talk by @SchmidhuberAI at the GTC 2021.

Important part when he cites Kunihiko Fukushima and his seminal work in the Neocognitron CNN architecture.

His work deserves a lot of recognition. Outstanding research in neural networks since the 70s!

[Luiz Felipe Vecchietti @lfvecchietti]

This article reminds me of the countless hours I spent in the studio manually editing speech and NLP datasets...

Months of invisible work for training 'old' seq-to-seq models.

This is a problem especially for languages without vastly available open source datasets.

[MIT Technology Review @techreview]

The 10 most cited AI data sets are riddled with label errors, according to a new study out of MIT, and it’s distorting our understanding of the field’s progress. https://t.co/dTXugbdpBa

[K League @kleague]

🎥 HIGHLIGHTS | Daejeon Hana Citizen 2-1 Gyeongnam FC

🔝 Daejeon move into pole position in @kleague 2 thanks to goals from Park In-hyeok and Pato.

📺 Watch 𝗟𝗜𝗩𝗘: https://t.co/A2QSGWlehT

#KLeague | #K리그 | #DHFCvGYE

[Paul Neat @NeatPaul]

What this means for #KLeague is that fans will still be allowed in stadiums. Seoul/Gyeonggi/Incheon still limited to 10%, as too are Busan, Jeonju (Jeonbuk) and Daejeon which are at Level 2.

https://t.co/PLK2x8LU9B

[K League United @KLeagueUnited]

Can Daejeon pick up a first home win of 2021?

#KLeague | #K리그
https://t.co/x6z2E8jA4j

[Luiz Felipe Vecchietti @lfvecchietti]

@MattSKim Start using overleaf!

[Luiz Felipe Vecchietti @lfvecchietti]

"Learners can benefit from this (evaluative feedback) by successfully inferring and internalizing a teacher's reward specification" (@Mark_Ho_ et al 2017)

A powerful sentence that summarizes a lot of the progress in Deep RL in 2020/2021.

[Luiz Felipe Vecchietti @lfvecchietti]

@ElTedao G2 feelings.

[Victor Escorcia @3scorciav]

@lfvecchietti Start by visiting

https://t.co/zz5yoEygRB

Get to know the activities to be hosted during the year. For example LXCV to be held in @CVPR #CVPR2021 during June.

https://t.co/8f5AcHvNlm

In sum, sign up, participate in the events, forums and connect with others 😊

[Luiz Felipe Vecchietti @lfvecchietti]

@3scorciav How to join?

[Luiz Felipe Vecchietti @lfvecchietti]

The work "Mutual Information State Intrinsic Control" by Rui Zhao, Yang Gao, @pabbeel, @vtresp, and Wei Xu is the most interesting Reinforcement Learning paper I read recently.

Amazing intuition. Simple concept. Good results.

https://t.co/Xg9wMcioLN