1) Changed!매주 화요일 오후 10시30분(Flexible) 논문 1개씩 리뷰 with some materials(like PPT)
2) Discord 채널 - https://discord.gg/vVjq8XwqPz
3) Issues : comment로 materials 첨부(link를 이용해서) + 한 주간 읽었던 논문 중에 흥미롭거나 소개해주고 싶은거 간단 요약과 링크
4) Coding Assignments : 1) Backpropagation with Numpy 구현 2) Transformer 구현 3)Backjun coding problem 4) ETC.
5) 발표시간 15분 + QNA 5분
6) If you want to presentation in English, It is possible!
sanha - Flow Network based Generative Models for Non-iterative Diverse Candidate Generationkeonwoo - INTEGER QUANTIZATION FOR DEEP LEARNING INFERENCE: PRINCIPLES AND EMPIRICAL EVALUATIONingeol - Multi-Task Deep Neural Networks for Natural Language Understanding, GPT2sanha - Decision Transformer via Reinforcement learninghoyeon - AutoAugment: Learning Augmentation Strategies from DataIngeol - Transformer XLjuwon- DCGANsanha - Masked AutoEncoder(MAE)keonwoo - Comprehensive survey about GNNhoyeon - Neural Machine Translation by Jointly learning to Align and TranslateIngeol - End-to-end Neural Coreference Resolutionjuwon- U-nethoyeon - Batch normalizationingeol - BERT: Pre-training of Deep Bidirectional Transformers forLanguage Understandingsanha - Improving Language Understanding by Generative Pre-Training (GPT1)juwon - Fully Convolution Networksanha - Visual prompting via image inpaintingkeonwoo - Variational Inference: A Review for Statisticianshoyeon - Sequence to Sequence Learning with Neural NetworksIngeol - Few-shot learningsanha - Generative Adversarial Imitation learning(GAIL)keonwoo - difussion modelhoyeon - Genarative adversarial networksanha - A study of inverse reinforcement learning and its implementation keonwoo - We need to know confidence of prediction on datasethoyeon - Autoencoding Variationa Bayessanha - Graphs Constraints and Search for the Abstraction and Reasoning Corpuskeonwoo - Attention is all you need!sanha - Stocastic prediction of multi agent interaction from partial observationskeonwoo - A Tutorial on Bayesian Optimizationhoyeon - Playing Atari with Deep Reinforcement Learningsanha - object centric learning with slot attention