본문 바로가기

연구실 공부

[논문] Learning Strides in Convolutional Neural Network https://openreview.net/forum?id=M752z9FKJP Learning Strides in Convolutional Neural Networks Convolutional neural networks typically contain several downsampling operators, such as strided convolutions or pooling layers, that progressively reduce the resolution of intermediate... openreview.net 해당 논문을 보고 작성했습니다. Abstract CNN은 일반적으로 strided convolution이나 pooling layer를 사용하여 downsampling을 진행합니다. 이.. 연구실 공부 2024. 2. 25.
[논문] Learning Transferable Visual Models from Language Supervision https://arxiv.org/abs/2103.00020 Learning Transferable Visual Models From Natural Language Supervision State-of-the-art computer vision systems are trained to predict a fixed set of predetermined object categories. This restricted form of supervision limits their generality and usability since additional labeled data is needed to specify any other visual co arxiv.org 해당 논문을 보고 작성했습니다. Abstract c.. 연구실 공부 2024. 2. 22.
[논문] LoRA: Low-Rank Adaptation of Large Language Models https://arxiv.org/abs/2106.09685 LoRA: Low-Rank Adaptation of Large Language Models An important paradigm of natural language processing consists of large-scale pre-training on general domain data and adaptation to particular tasks or domains. As we pre-train larger models, full fine-tuning, which retrains all model parameters, becomes le arxiv.org 해당 논문을 보고 작성했습니다. Abstract 자연어 처리 분야에서 중요한 패러다임.. 연구실 공부 2024. 2. 20.
[논문] Deep Residual Learning for Image Recognition https://arxiv.org/abs/1512.03385 Deep Residual Learning for Image Recognition Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with arxiv.org 해당 논문을 보고 작성했습니다. Abstract 더 깊은 neural network은 train.. 연구실 공부 2024. 2. 19.
[논문] Generative Adversarial Net https://arxiv.org/pdf/1406.2661.pdf 해당 논문을 보고 작성했습니다. Abstract adversarial process를 통해 생성 모델을 추정하는 새로운 framework를 제언합니다. 이는 2개의 모델을 동시에 학습합니다. 데이터 분포를 capture하는 생성 모델 G(생성자), G에서 생성된 sample인지 train data에서 온 sample인지에 대한 확률을 추정하는 discriminative model D입니다(판별자). G는 D가 mistake할 확률을 maximize하도록 학습합니다(즉, 본인이 만든 sample을 train data에서 왔다고 판별하도록). 이는 minimax two-player game과 같은 형태입니다. 임의의 함수 G와 D의 space에.. 연구실 공부 2024. 2. 17.
[논문] Large-Scale Self-Supervised Speech Representation Learning for Automatic Speaker Verification https://arxiv.org/abs/2110.05777 Large-scale Self-Supervised Speech Representation Learning for Automatic Speaker Verification The speech representations learned from large-scale unlabeled data have shown better generalizability than those from supervised learning and thus attract a lot of interest to be applied for various downstream tasks. In this paper, we explore the limits of arxiv.org 해당 논.. 연구실 공부 2024. 2. 8.
[논문] ResNeXt and Res2Net Structures for Speaker verification https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9383531 ResNeXt and Res2Net Structures for Speaker Verification The ResNet-based architecture has been widely adopted to extract speaker embeddings for text-independent speaker verification systems. By introducing the residual connections to the CNN and standardizing the residual blocks, the ResNet structure is capable ieeexplore.ieee.org 해당 논.. 연구실 공부 2024. 2. 7.
[논문] ECAPA-TDNN: Emphasized Channel Attention, Propagation and Aggregation in TDNN Based Speaker Verification https://arxiv.org/abs/2005.07143 ECAPA-TDNN: Emphasized Channel Attention, Propagation and Aggregation in TDNN Based Speaker Verification Current speaker verification techniques rely on a neural network to extract speaker representations. The successful x-vector architecture is a Time Delay Neural Network (TDNN) that applies statistics pooling to project variable-length utterances into fixed arx.. 연구실 공부 2024. 2. 6.
728x90

티스토리툴바