no code implementations • 14 Apr 2024 • Xin-Chun Li, Shaoming Song, Yinchuan Li, Bingshuai Li, Yunfeng Shao, Yang Yang, De-Chuan Zhan
For better model personalization, we point out that the hard-won personalized models are not well exploited and propose "inherited private model" to store the personalization experience.
no code implementations • 4 Apr 2024 • Kaichen Huang, Hai-Hang Sun, Shenghua Wan, Minghao Shao, Shuai Feng, Le Gan, De-Chuan Zhan
Imitating skills from low-quality datasets, such as sub-optimal demonstrations and observations with distractors, is common in real-world applications.
no code implementations • 4 Apr 2024 • Kaichen Huang, Minghao Shao, Shenghua Wan, Hai-Hang Sun, Shuai Feng, Le Gan, De-Chuan Zhan
In many real-world visual Imitation Learning (IL) scenarios, there is a misalignment between the agent's and the expert's perspectives, which might lead to the failure of imitation.
no code implementations • 20 Mar 2024 • Chao Yi, De-Chuan Zhan, Han-Jia Ye
It then uses this matrix to transfer useful statistics of VLMs from open-source datasets to the target dataset for bridging those two gaps and enhancing the VLM's capacity estimation for VLM selection.
1 code implementation • 18 Mar 2024 • Da-Wei Zhou, Hai-Long Sun, Han-Jia Ye, De-Chuan Zhan
Despite the strong performance of Pre-Trained Models (PTMs) in CIL, a critical issue persists: learning new classes often results in the overwriting of old ones.
no code implementations • 15 Mar 2024 • Yucen Wang, Shenghua Wan, Le Gan, Shuai Feng, De-Chuan Zhan
Model-based methods have significantly contributed to distinguishing task-irrelevant distractors for visual control.
1 code implementation • 29 Jan 2024 • Da-Wei Zhou, Hai-Long Sun, Jingyi Ning, Han-Jia Ye, De-Chuan Zhan
Nowadays, real-world applications often face streaming data, which requires the learning system to absorb new knowledge as data evolves.
no code implementations • 27 Dec 2023 • Lan Li, Bowen Tao, Lu Han, De-Chuan Zhan, Han-Jia Ye
Differing from traditional semi-supervised learning, class-imbalanced semi-supervised learning presents two distinct challenges: (1) The imbalanced distribution of training samples leads to model bias towards certain classes, and (2) the distribution of unlabeled samples is unknown and potentially distinct from that of labeled samples, which further contributes to class bias in the pseudo-labels during training.
no code implementations • 15 Dec 2023 • Bowen Tao, Lan Li, Xin-Chun Li, De-Chuan Zhan
For each pseudo-labeled sample, we select positive and negative samples from labeled data instead of unlabeled data to compute contrastive loss.
1 code implementation • NeurIPS 2023 • Qi-Wei Wang, Da-Wei Zhou, Yi-Kai Zhang, De-Chuan Zhan, Han-Jia Ye
In this Few-Shot Class-Incremental Learning (FSCIL) scenario, existing methods either introduce extra learnable components or rely on a frozen feature extractor to mitigate catastrophic forgetting and overfitting problems.
1 code implementation • 30 Nov 2023 • Lu Han, Xu-Yang Chen, Han-Jia Ye, De-Chuan Zhan
This achievement not only underscores the effectiveness of our methodologies but also establishes a new standard for deep learning applications in weather forecasting.
no code implementations • 31 Oct 2023 • Han-Jia Ye, Qi-Le Zhou, De-Chuan Zhan
Tabular data is prevalent across various machine learning domains.
no code implementations • 23 Oct 2023 • Qi-Le Zhou, Han-Jia Ye, Le-Ye Wang, De-Chuan Zhan
Fine-tuning a pre-trained deep neural network has become a successful paradigm in various machine learning tasks.
1 code implementation • 26 Sep 2023 • Songli Wu, Liang Du, Jia-Qi Yang, Yuai Wang, De-Chuan Zhan, Shuang Zhao, Zixun Sun
Click-through rate (CTR) prediction is a critical task in recommendation systems, serving as the ultimate filtering step to sort items for a user.
1 code implementation • 13 Sep 2023 • Hai-Long Sun, Da-Wei Zhou, Han-Jia Ye, De-Chuan Zhan
While traditional machine learning can effectively tackle a wide range of problems, it primarily operates within a closed-world setting, which presents limitations when dealing with streaming data.
1 code implementation • 17 Aug 2023 • Yi-Kai Zhang, Lu Ren, Chao Yi, Qi-Wei Wang, De-Chuan Zhan, Han-Jia Ye
The rapid expansion of foundation pre-trained models and their fine-tuned counterparts has significantly contributed to the advancement of machine learning.
no code implementations • 14 Jul 2023 • Qi-Wei Wang, Hongyu Lu, Yu Chen, Da-Wei Zhou, De-Chuan Zhan, Ming Chen, Han-Jia Ye
The Click-Through Rate (CTR) prediction task is critical in industrial recommender systems, where models are usually deployed on dynamic streaming data in practical applications.
no code implementations • 19 Jun 2023 • Shenghua Wan, Yucen Wang, Minghao Shao, Ruying Chen, De-Chuan Zhan
Model-based imitation learning (MBIL) is a popular reinforcement learning method that improves sample efficiency on high-dimension input sources, such as images and videos.
no code implementations • 8 Jun 2023 • Jia-Qi Yang, Chenglei Dai, OU Dan, Ju Huang, De-Chuan Zhan, Qingwen Liu, Xiaoyi Zeng, Yang Yang
To this end, we propose a recommendation-aware image pre-training method that can learn visual features from user click histories.
1 code implementation • 30 May 2023 • Jia-Qi Yang, Yucheng Xu, Jia-Lei Shen, Kebin Fan, De-Chuan Zhan, Yang Yang
These obstacles prevent AI researchers from developing specialized methods for scientific designs.
no code implementations • 30 May 2023 • Da-Wei Zhou, Yuanhan Zhang, Jingyi Ning, Han-Jia Ye, De-Chuan Zhan, Ziwei Liu
While traditional CIL methods focus on visual information to grasp core features, recent advances in Vision-Language Models (VLM) have shown promising capabilities in learning generalizable representations with the aid of textual information.
1 code implementation • 7 May 2023 • Xin-Chun Li, Yang Yang, De-Chuan Zhan
We propose a novel learning paradigm named transductive federated learning (TFL) to simultaneously consider the structural information of the to-be-inferred data.
1 code implementation • 14 Apr 2023 • Bowen Zheng, Da-Wei Zhou, Han-Jia Ye, De-Chuan Zhan
In this paper, we encourage the model to preserve more local information as the training procedure goes on and devise a Locality-Preserved Attention (LPA) layer to emphasize the importance of local features.
1 code implementation • 11 Apr 2023 • Lu Han, Han-Jia Ye, De-Chuan Zhan
Our results conclude that the CD approach has higher capacity but often lacks robustness to accurately predict distributionally drifted time series.
2 code implementations • 13 Mar 2023 • Da-Wei Zhou, Han-Jia Ye, De-Chuan Zhan, Ziwei Liu
ADAM is a general framework that can be orthogonally combined with any parameter-efficient tuning method, which holds the advantages of PTM's generalizability and adapted model's adaptivity.
2 code implementations • 7 Feb 2023 • Da-Wei Zhou, Qi-Wei Wang, Zhi-Hong Qi, Han-Jia Ye, De-Chuan Zhan, Ziwei Liu
Deep models, e. g., CNNs and Vision Transformers, have achieved impressive achievements in many vision tasks in the closed world.
no code implementations • 15 Jan 2023 • Lu Han, Han-Jia Ye, De-Chuan Zhan
Based on the findings, we propose to improve PL in class-mismatched SSL with two components -- Re-balanced Pseudo-Labeling (RPL) and Semantic Exploration Clustering (SEC).
1 code implementation • 5 Jan 2023 • Shaowei Zhang, Jiahan Cao, Lei Yuan, Yang Yu, De-Chuan Zhan
In cooperative multi-agent reinforcement learning (CMARL), it is critical for agents to achieve a balance between self-exploration and team collaboration.
1 code implementation • CVPR 2023 • Yi-Kai Zhang, Qi-Wei Wang, De-Chuan Zhan, Han-Jia Ye
When a dataset is biased, i. e., most samples have attributes spuriously correlated with the target label, a Deep Neural Network (DNN) is prone to make predictions by the "unintended" attribute, especially if it is easier to learn.
no code implementations • 10 Oct 2022 • Xin-Chun Li, Wen-Shu Fan, Shaoming Song, Yinchuan Li, Bingshuai Li, Yunfeng Shao, De-Chuan Zhan
Complex teachers tend to be over-confident and traditional temperature scaling limits the efficacy of {\it class discriminability}, resulting in less discriminative wrong class probabilities.
1 code implementation • 17 Jun 2022 • Xin-Chun Li, Jin-Lin Tang, Shaoming Song, Bingshuai Li, Yinchuan Li, Yunfeng Shao, Le Gan, De-Chuan Zhan
Federated KWS (FedKWS) could serve as a solution without directly sharing users' data.
1 code implementation • 1 Jun 2022 • Jia-Qi Yang, De-Chuan Zhan
We propose a generalized delayed feedback model (GDFM) that unifies both post-click behaviors and early conversions as stochastic post-click information, which could be utilized to train GDFM in a streaming manner efficiently.
1 code implementation • 1 Jun 2022 • Lu Han, Han-Jia Ye, De-Chuan Zhan
Self-supervised learning aims to learn a embedding space where semantically similar samples are close.
2 code implementations • 26 May 2022 • Da-Wei Zhou, Qi-Wei Wang, Han-Jia Ye, De-Chuan Zhan
We find that when counting the model size into the total budget and comparing methods with aligned memory size, saving models do not consistently work, especially for the case with limited memory budgets.
1 code implementation • 4 May 2022 • Han-Jia Ye, Su Lu, De-Chuan Zhan
Instead of enforcing the teacher to work on the same task as the student, we borrow the knowledge from a teacher trained from a general label space -- in this "Generalized Knowledge Distillation (GKD)", the classes of the teacher and the student may be the same, completely different, or partially overlapped.
no code implementations • 25 Apr 2022 • Su Lu, Han-Jia Ye, De-Chuan Zhan
Our method reuses cross-task knowledge from a distinct label space and efficiently assesses teachers without enumerating the model repository.
2 code implementations • 10 Apr 2022 • Fu-Yun Wang, Da-Wei Zhou, Han-Jia Ye, De-Chuan Zhan
The ability to learn new concepts continually is necessary in this ever-changing world.
Ranked #1 on Incremental Learning on ImageNet100 - 20 steps
1 code implementation • CVPR 2022 • Han-Jia Ye, Yi Shi, De-Chuan Zhan
To this end, we introduce a novel evaluation criterion by predicting the comparison's correctness after assigning the learned embeddings to their optimal conditions, which measures how much WS-CSL could cover latent semantics as the supervised model.
1 code implementation • 31 Mar 2022 • Da-Wei Zhou, Han-Jia Ye, Liang Ma, Di Xie, ShiLiang Pu, De-Chuan Zhan
In this work, we propose a new paradigm for FSCIL based on meta-learning by LearnIng Multi-phase Incremental Tasks (LIMIT), which synthesizes fake FSCIL tasks from the base dataset.
Ranked #5 on Few-Shot Class-Incremental Learning on CIFAR-100
no code implementations • CVPR 2022 • Xin-Chun Li, Yi-Chu Xu, Shaoming Song, Bingshuai Li, Yinchuan Li, Yunfeng Shao, De-Chuan Zhan
The permutation invariance property of neural networks and the non-i. i. d.
1 code implementation • CVPR 2022 • Da-Wei Zhou, Fu-Yun Wang, Han-Jia Ye, Liang Ma, ShiLiang Pu, De-Chuan Zhan
Forward compatibility requires future new classes to be easily incorporated into the current model based on the current stage data, and we seek to realize it by reserving embedding space for future new classes.
Ranked #4 on Few-Shot Class-Incremental Learning on CIFAR-100
1 code implementation • 23 Dec 2021 • Da-Wei Zhou, Fu-Yun Wang, Han-Jia Ye, De-Chuan Zhan
Traditional machine learning systems are deployed under the closed-world setting, which requires the entire training data before the offline training process.
1 code implementation • 7 Dec 2021 • Jia-Qi Yang, Ke-Bin Fan, Hao Ma, De-Chuan Zhan
We also define a sample-wise weight, which can be used in the maximum weighted likelihood estimation of an inverse model based on a cINN.
no code implementations • 26 Sep 2021 • Jiahan Cao, Lei Yuan, Jianhao Wang, Shaowei Zhang, Chongjie Zhang, Yang Yu, De-Chuan Zhan
During long-time observations, agents can build \textit{awareness} for teammates to alleviate the problem of partial observability.
2 code implementations • 27 Jul 2021 • Da-Wei Zhou, Han-Jia Ye, De-Chuan Zhan
As a result, we propose CO-transport for class Incremental Learning (COIL), which learns to relate across incremental tasks with the class-wise semantic relationship.
no code implementations • 26 Jul 2021 • Xin-Chun Li, Le Gan, De-Chuan Zhan, Yunfeng Shao, Bingshuai Li, Shaoming Song
We advocate the proposed methods could serve as a preliminary try to explore where to privatize for a novel non-iid scene.
no code implementations • 26 Jul 2021 • Xin-Chun Li, Lan Li, De-Chuan Zhan, Yunfeng Shao, Bingshuai Li, Shaoming Song
Automatically mining sentiment tendency contained in natural language is a fundamental research to some artificial intelligent applications, where solutions alternate with challenges.
1 code implementation • 1 Jul 2021 • Han-Jia Ye, Lu Ming, De-Chuan Zhan, Wei-Lun Chao
Many existing works take the meta-learning approach, constructing a few-shot learner that can learn from few-shot examples to generate a classifier.
1 code implementation • 15 Jun 2021 • Han-Jia Ye, Da-Wei Zhou, Lanqing Hong, Zhenguo Li, Xiu-Shen Wei, De-Chuan Zhan
To this end, we propose Learning to Decompose Network (LeadNet) to contextualize the meta-learned ``support-to-target'' strategy, leveraging the context of instances with one or mixed latent attributes in a support set.
no code implementations • 17 Apr 2021 • Yang Yang, Zhao-Yang Fu, De-Chuan Zhan, Zhi-Bin Liu, Yuan Jiang
Moreover, we introduce the extrinsic unlabeled multi-modal multi-instance data, and propose the M3DNS, which considers the instance-level auto-encoder for single modality and modified bag-level optimal transport to strengthen the consistency among modalities.
1 code implementation • NeurIPS 2021 • Su Lu, Han-Jia Ye, Le Gan, De-Chuan Zhan
Different from $\mathcal{S}$/$\mathcal{Q}$ protocol, we can also evaluate a task-specific solver by comparing it to a target model $\mathcal{T}$, which is the optimal model for this task or a model that behaves well enough on this task ($\mathcal{S}$/$\mathcal{T}$ protocol).
no code implementations • 8 Apr 2021 • Su Lu, Han-Jia Ye, De-Chuan Zhan
In detail, given two videos, we sample segments from them and cast the calculation of their distance as an optimal transport problem between two segment sequences.
no code implementations • ICCV 2021 • Han-Jia Ye, De-Chuan Zhan, Wei-Lun Chao
To correct these wrong predictions, the neural network then must focus on pushing features of minor class data across the decision boundaries between major and minor classes, leading to much larger gradients for features of minor classes.
1 code implementation • CVPR 2021 • Da-Wei Zhou, Han-Jia Ye, De-Chuan Zhan
To this end, we proposed to learn PlaceholdeRs for Open-SEt Recognition (Proser), which prepares for the unknown classes by allocating placeholders for both data and classifier.
1 code implementation • 6 Dec 2020 • Jia-Qi Yang, Xiang Li, Shuguang Han, Tao Zhuang, De-Chuan Zhan, Xiaoyi Zeng, Bin Tong
To strike a balance in this trade-off, we propose Elapsed-Time Sampling Delayed Feedback Model (ES-DFM), which models the relationship between the observed conversion distribution and the true conversion distribution.
1 code implementation • 30 Nov 2020 • Han-Jia Ye, Lu Han, De-Chuan Zhan
Meta-learning has become a practical approach towards few-shot image classification, where "a strategy to learn a classifier" is meta-learned on labeled base classes and can be applied to tasks with novel classes.
Unsupervised Few-Shot Image Classification Unsupervised Few-Shot Learning
no code implementations • 3 Feb 2020 • Wei-Lun Chao, Han-Jia Ye, De-Chuan Zhan, Mark Campbell, Kilian Q. Weinberger
Recent years have witnessed an abundance of new publications and approaches on meta-learning.
1 code implementation • 6 Jan 2020 • Han-Jia Ye, Hong-You Chen, De-Chuan Zhan, Wei-Lun Chao
Classifiers trained with class-imbalanced data are known to perform poorly on test data of the "minor" classes, of which we have insufficient training data.
1 code implementation • 7 Jun 2019 • Han-Jia Ye, Hexiang Hu, De-Chuan Zhan
In this paper, we investigate the problem of generalized few-shot learning (GFSL) -- a model during the deployment is required to learn about tail categories with few shots and simultaneously classify the head classes.
5 code implementations • CVPR 2020 • Han-Jia Ye, Hexiang Hu, De-Chuan Zhan, Fei Sha
Many few-shot learning methods address this challenge by learning an instance embedding function from seen classes and apply the function to instances from unseen classes with limited labels.
no code implementations • ICML 2018 • Han-Jia Ye, De-Chuan Zhan, Yuan Jiang, Zhi-Hua Zhou
On the way to the robust learner for real-world applications, there are still great challenges, including considering unknown environments with limited data.
no code implementations • NeurIPS 2016 • Han-Jia Ye, De-Chuan Zhan, Xue-Min Si, Yuan Jiang, Zhi-Hua Zhou
In UM2L, a type of combination operator is introduced for distance characterization from multiple perspectives, and thus can introduce flexibilities for representing and utilizing both spatial and semantic linkages.