# MSP-AI Meeting

> [name=周惶振]

> [time=Fri, Apr 2, 2021 5:47 AM]

# [Meeting Link](https://www.google.com/url?q=https://teams.microsoft.com/l/meetup-join/19%253ameeting_NDU2NTFiMjgtOWQ5Yy00MjdiLTgzOTMtNThkYzkxNjU3ODk2%2540thread.v2/0?context%3D%257b%2522Tid%2522%253a%25228d281d1d-9c4d-4bf7-b16e-032d15de9f6c%2522%252c%2522Oid%2522%253a%25225a25d468-070a-47f7-b497-298163e08cca%2522%257d&sa=D&source=calendar&ust=1617745445611000&usg=AOvVaw2goKmvb2H_31mg9o34rjvM)

- [Server Setup](https://docs.google.com/document/d/13sM-6Y-o_03JxaEzu0fiWIZXKzKYWzB8JJGFFz5_vHA/edit)

- [Meeting](https://teams.microsoft.com/l/meetup-join/19%3ameeting_NGFiZWVmMGItNWEyYS00MTFlLWIzMTgtZjkyZTNlMzYyZjRm%40thread.v2/0?context=%7b%22Tid%22%3a%228d281d1d-9c4d-4bf7-b16e-032d15de9f6c%22%2c%22Oid%22%3a%220637d254-c6ff-4eb9-bd61-a2997b9488bc%22%7d)

# Paper
## 2021~2022 Fall
- [] [12/02 Labelling unlabelled videos from scratch with multi-modal self-supervision](https://arxiv.org/pdf/2006.13662.pdf)

- [] [11/04 Online Continual Learning from Imbalanced Data](http://proceedings.mlr.press/v119/chrysakis20a.html)
-

- [] [10/14 Self-supervised Co-training for Video Representation Learning]
- https://www.robots.ox.ac.uk/~vgg/research/CoCLR/
- [] [10/07]

- [] [09/30 Automatic Shortcut Removal for Self-Supervised Representation Learning](https://arxiv.org/pdf/2002.08822.pdf)
- [ ] Summary
- [ ] Contribtuion
- [ ] Strength
* Novelty
* Contribtuion
* I can use
*
- [ ] Weakness
* What needs more

- [ ] [09/09 Audio-Visual Instance Discrimination with Cross-Modal Agreement](https://openaccess.thecvf.com/content/CVPR2021/papers/Morgado_Audio-Visual_Instance_Discrimination_with_Cross-Modal_Agreement_CVPR_2021_paper.pdf)
- [ ] Strength
* Novelty
* Contribtuion
* I can use
*
- [ ] Weakness
* What needs more
*

- [ ]

# 2020~2021 Spring
- [ ] [08/05 CVPR' 2018 End-to-End Incremental Learning](https://openaccess.thecvf.com/content_ECCV_2018/papers/Francisco_M._Castro_End-to-End_Incremental_Learning_ECCV_2018_paper.pdf)
- [ ] Background
- [ ] Novelty
- [ ] Distilla loss

- [ ] [07/29]
- [ ] [07/01 CVPR 2020': S3VAE: Self-Supervised Sequential VAE for Representation Disentanglement and Data Generation](https://arxiv.org/pdf/2005.11437.pdf)
* Background Knowledge:
* Time-varying
* motion
* Time-invariant
* objects
* timbre of speakers
* Benefits of learning disentangled representations
* More explainable
* Easier and more efficient to manipulate data gen- eration
* Disentangled Sequential Data Generation
* Self-Supervised Learning
* Intrinsic labels
* Auxiliary data
* Off-the-shelf tools
* Variational inference
* Main Challenges:
* Label is too expensive
* Unsurvised learning is hard to perform well
* Main Novelty:
* Leverage supervisory signals from intrinsic la- bels to regularize the static representation and off-the-shelf tools to regularize the dynamic representation
* Model both the prior and the posterior of zt by recurrent models independently -> consistent dynamic information in synthetic sequences
* Ensure full disentanglement of zf and zt by posterior factorization
* Contributions:
* Fewer works have explored representation disentanglement for sequential data generation
* Proposed Approach:
*

- [ ] [06/24 NeurIPS 2020': Causal Intervention for Weakly-Supervised Semantic Segmentation](https://arxiv.org/pdf/2009.12547.pdf)
* Background Knowledge:
* Classification Activation Map (CAM)
* Pseudo-Masks
* associates non-causal but positively correlated pixels to labels
* disassociates causal but negatively correlated ones
* multi-label soft-margin loss: [A mixed objective optimization network for the recognition of facial attributes](https://www.researchgate.net/profile/Manuel-Guenther/publication/301838380_MOON_A_Mixed_Objective_Optimization_Network_for_the_Recognition_of_Facial_Attributes/links/59fca36faca272347a220b97/MOON-A-Mixed-Objective-Optimization-Network-for-the-Recognition-of-Facial-Attributes.pdf)
* Graphical model vs Graph model
* [Graphical model](https://zh.wikipedia.org/zh-tw/%E5%9C%96%E6%A8%A1%E5%BC%8F)
* Main Challenges:

* Main Novelty and Contributions:
* Context Adjustment (CONTA)

* Proposed Approach:
- [ ] [06/17 WACV 2021': Class-incremental Learning via Deep Model Consolidation](https://openaccess.thecvf.com/content_WACV_2020/papers/Zhang_Class-incremental_Learning_via_Deep_Model_Consolidation_WACV_2020_paper.pdf)
* Main Challenges:

* Main Novelty and Contributions:

* Proposed Approach:
* Step1: To train a (t − s)-class classifier using training data Dnew, which we refer as the new model $f_{new}(x; Θnew)$
* Step2: To consolidate the old model and the new model (Deep Model Consolidation (DMC) for image classification)

* Similarity: Federated learning






- [ ] [06/10 CVPR 2019': Learning a Unified Classifier Incrementally via Rebalancing](https://openaccess.thecvf.com/content_CVPR_2019/papers/Hou_Learning_a_Unified_Classifier_Incrementally_via_Rebalancing_CVPR_2019_paper.pdf)
* Main Challenges:
* Incremental Learning:
* Preserve parameters of the orignal model
* Catastropic forgetting

* Main Novelty and Contributions:
* For catastropic forgetting:
* Cosine-norm
* 
* Less-forget constraint
* 
* Inter-class separation
* 

- [ ] [06/03 ICLR 2021': RETHINKING THE ROLE OF GRADIENT-BASED ATTRIBUTION METHODS FOR MODEL INTERPRETABILITY](https://openreview.net/pdf/b29e31cf78e011a59f9e49950670211d4c516b00.pdf)
* Simpler:
* [Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps](https://arxiv.org/pdf/1312.6034.pdf)
*
* Interpretability vs prior
*
* Prior 、Posterior 和Likelihood

- [ ][05/28 On the uncertainty of self-supervised monocular depth estimation](https://openaccess.thecvf.com/content_CVPR_2020/papers/Poggi_On_the_Uncertainty_of_Self-Supervised_Monocular_Depth_Estimation_CVPR_2020_paper.pdf)
* Uncertainty modeling methods (how reliable on your predictions)
- Mean and Variance
- 
- [Cosine Scheduling](https://arxiv.org/pdf/1608.03983v5.pdf)
- Uncertainty metric:
- Area Under the Sparsification Error (AUSE)
- Area Under the Random Gain (AURU) (higher is better)

- [ ] [Attention-Based Models for Speech Recognition](https://arxiv.org/pdf/1506.07503.pdf)
- [ ] Adaptive Learning Rate Algorithm
- [ ] Use sigmoid to apply on attention
- [ ] Optimizer:
- [ ] Older: AdaDelta
- [ ] Adam
- [ ]
- [ ] [中文](https://zhuanlan.zhihu.com/p/54172131)
- [ ] 其他參考資料:
- [ ] [Neural Turing Machines](https://arxiv.org/pdf/1410.5401v2.pdf)
- [ ] [李宏毅老師](http://speech.ee.ntu.edu.tw/~tlkagk/courses/DLHLP20/ASR%20(v12).pdf)

- [ ] [Teaching a GAN what not to learn](https://arxiv.org/pdf/2010.15639.pdf)
- [ ] [知乎](https://zhuanlan.zhihu.com/p/279113851)
- [ ] [Github](https://githubmemory.com/repo/DarthSid95/RumiGANs)
- [ ] Frechet inception distance (FID) socres

- [ ] [紀錄Miro](https://miro.com/app/)

- [ ] [Bootstrap Your Own Latent: A New Approach to Self-Supervised Learning]
- [ ] Novelty
- [ ] No negative smaples
- [ ] Self-supervised
- [ ] Bootstrap Method，Bootstrapping

- [] [Deep Adversarial Metric Learning]
- 
- [ ][強baseline-A Metric Learning Reality Check](https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123700681.pdf)
- [ ][Metric Learning Survey](https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123700681.pdf)
- [ ][A tutorial on distance metric learning: Mathematical foundations,algorithms, experimental analysis, prospects and challenges](https://reader.elsevier.com/reader/sd/pii/S0925231220312777?token=959EAF77DFB880FF5E4DBE3A3B0BF1D85D7DA8CFF6E1A2A51741860B9A4BA6FD724534EE77502AE4B943726995178C9F&originRegion=us-east-1&originCreation=20210422225727)
- [ ][Learning to Self-Train for Semi-Supervised Few-Shot Classification](https://proceedings.neurips.cc/paper/2019/file/bf25356fd2a6e038f1a3a59c26687e80-Paper.pdf)
- Use non-linear distance
- Linear Mahlanobis distance
- Large margin nearest neighbor
- [] [Novelty]
- Generate synthetic negative examples
- Goal
- 
- 
- 
- Other Refs.
- [Introduce](http://contrib.scikit-learn.org/metric-learn/introduction.html)
- [Survey](file:///C:/Users/User/Desktop/symmetry-11-01066-v2.pdf)
- [Deep Metric Learning with Graph Consistency](https://www.aaai.org/AAAI21Papers/AAAI-1310.ChenB.pdf)
- [FOCAL: EFFICIENT FULLY-OFFLINE METAREINFORCEMENT LEARNING VIA DISTANCE METRIC LEARNING AND BEHAVIOR REGULARIZATION](https://openreview.net/attachment?id=8cpHIfgY4Dj&name=pdf)
- [UNSUPERVISED DOMAIN ADAPTATION FOR DISTANCE METRIC LEARNING](https://openreview.net/pdf?id=BklhAj09K7)
- [Fewer is More: A Deep Graph Metric Learning
Perspective Using Fewer Proxies](https://papers.nips.cc/paper/2020/file/ce016f59ecc2366a43e1c96a4774d167-Paper.pdf)

- [ ] [RANet: Region Attention Network for Semantic Segmentation](https://papers.nips.cc/paper/2020/file/9fe8593a8a330607d76796b35c64c600-Paper.pdf)
- [Supplemental](https://papers.nips.cc/paper/2020/file/9fe8593a8a330607d76796b35c64c600-Supplemental.pdf)
- Main Goal:
- Classifying each pixel in an image to a class label
- Important task for understanding images
- Main idea:
- 
- Method:
- 
- Region Construction Block
- 
- 
- 
- $G_{ij}$ = -1,0,1

- [ ] [Stand-Alone Self-Attention in Vision Models](https://papers.nips.cc/paper/2019/file/3416a75f4cea9109507cacd8e2f2aefc-Paper.pdf)
- [Another Paper:Attention Augmented Convolutional Networks](https://arxiv.org/pdf/1904.09925.pdf)
- [ ] Novelty:
- Replace CNN with attention
- Specialized Layer for images
- Spatially localized features
- Translation invariance
- Less memory and computations
- Create a stand-alone model using only self-attetnion
- 
- 

- [ ] Results:
- [ ] To show the efficiency of models
-