1 code implementation • 12 Jan 2024 • Chandler Timm Doloriel, Ngai-Man Cheung
We study spatial and frequency domain masking in training deepfake detectors.
1 code implementation • 26 Jul 2023 • Milad Abdollahzadeh, Touba Malekzadeh, Christopher T. H. Teo, Keshigeyan Chandrasegaran, Guimeng Liu, Ngai-Man Cheung
In machine learning, generative modeling aims to learn to generate new data statistically similar to the training data distribution.
no code implementations • 4 Jul 2023 • Yunqing Zhao, Keshigeyan Chandrasegaran, Milad Abdollahzadeh, Chao Du, Tianyu Pang, Ruoteng Li, Henghui Ding, Ngai-Man Cheung
However, a major limitation of existing methods is that their knowledge preserving criteria consider only source domain/task and fail to consider target domain/adaptation in selecting source knowledge, casting doubt on their suitability for setups of different proximity between source and target domain.
1 code implementation • NeurIPS 2023 • Yunqing Zhao, Tianyu Pang, Chao Du, Xiao Yang, Chongxuan Li, Ngai-Man Cheung, Min Lin
Large vision-language models (VLMs) such as GPT-4 have achieved unprecedented performance in response generation, especially with visual inputs, enabling more creative and adaptable interaction than large language models such as ChatGPT.
1 code implementation • CVPR 2023 • Yunqing Zhao, Chao Du, Milad Abdollahzadeh, Tianyu Pang, Min Lin, Shuicheng Yan, Ngai-Man Cheung
To this end, we propose knowledge truncation to mitigate this issue in FSIG, which is a complementary operation to knowledge preservation and is implemented by a lightweight pruning-based method.
1 code implementation • CVPR 2023 • Ngoc-Bao Nguyen, Keshigeyan Chandrasegaran, Milad Abdollahzadeh, Ngai-Man Cheung
Recently, several algorithms for MI have been proposed to improve the attack performance.
1 code implementation • 17 Mar 2023 • Yunqing Zhao, Tianyu Pang, Chao Du, Xiao Yang, Ngai-Man Cheung, Min Lin
Diffusion models (DMs) have demonstrated advantageous potential on generative tasks.
1 code implementation • 2 Dec 2022 • Christopher TH Teo, Milad Abdollahzadeh, Ngai-Man Cheung
We find that our fairTL can learn expressive sample generation during pre-training, thanks to the large (biased) dataset.
2 code implementations • 29 Oct 2022 • Yunqing Zhao, Keshigeyan Chandrasegaran, Milad Abdollahzadeh, Ngai-Man Cheung
However, a major limitation of existing methods is that their knowledge preserving criteria consider only source domain/source task, and they fail to consider target domain/adaptation task in selecting source model's knowledge, casting doubt on their suitability for setups of different proximity between source and target domain.
Ranked #1 on 10-shot image generation on Babies
1 code implementation • 24 Aug 2022 • Keshigeyan Chandrasegaran, Ngoc-Trung Tran, Alexander Binder, Ngai-Man Cheung
Visual counterfeits are increasingly causing an existential conundrum in mainstream media with rapid evolution in neural image synthesis methods.
1 code implementation • 23 Aug 2022 • Yunqing Zhao, Ngai-Man Cheung
This improved generalization motivates us to study BAN for DG-FSC, and we show that BAN is promising to address the domain shift encountered in DG-FSC.
no code implementations • 31 Jul 2022 • Evan Ehrenberg, Kleovoulos Leo Tsourides, Hossein Nejati, Ngai-Man Cheung, Pawan Sinha
In the domain of face recognition, there exists a puzzling timing discrepancy between results from macaque neurophysiology on the one hand and human electrophysiology on the other.
1 code implementation • 29 Jun 2022 • Keshigeyan Chandrasegaran, Ngoc-Trung Tran, Yunqing Zhao, Ngai-Man Cheung
Critically, there is no effort to understand and resolve these contradictory findings, leaving the primal question -- to smooth or not to smooth a teacher network?
no code implementations • CVPR 2022 • Yunqing Zhao, Henghui Ding, Houjing Huang, Ngai-Man Cheung
Informed by our analysis and to slow down the diversity degradation of the target generator during adaptation, our second contribution proposes to apply mutual information (MI) maximization to retain the source domain's rich multi-level diversity information in the target domain generator.
Ranked #2 on 10-shot image generation on Babies
1 code implementation • 16 Dec 2021 • Thilini Cooray, Ngai-Man Cheung
Hence, this work proposes a new principle for unsupervised graph representation learning: Graph-wise Common latent Factor EXtraction (GCFX).
no code implementations • 13 Dec 2021 • Tuan Hoang, Thanh-Toan Do, Tam V. Nguyen, Ngai-Man Cheung
First, to learn informative representations that can preserve both intra- and inter-modal similarities, we leverage the recent advances in estimating variational lower-bound of MI to maximize the MI between the binary representations and input features and between binary representations of different modalities.
1 code implementation • NeurIPS 2021 • Milad Abdollahzadeh, Touba Malekzadeh, Ngai-Man Cheung
Second, inspired by hard parameter sharing in multi-task learning and a new interpretation of related work, we propose a new multimodal meta-learner that outperforms existing work by considerable margins.
no code implementations • 24 Oct 2021 • Yi Xiang Marcus Tan, Penny Chong, Jiamei Sun, Ngai-Man Cheung, Yuval Elovici, Alexander Binder
In this work, we aim to close this gap by studying a conceptually simple approach to defend few-shot classifiers against adversarial attacks.
no code implementations • 29 Sep 2021 • Keshigeyan Chandrasegaran, Ngoc-Trung Tran, Yunqing Zhao, Ngai-Man Cheung
On the contrary, Shen et al. [2] claim that LS enlarges the distance between semantically similar classes; therefore a LS-trained teacher is compatible with KD.
1 code implementation • 16 Jul 2021 • Christopher T. H Teo, Ngai-Man Cheung
Deep generative models have made much progress in improving training stability and quality of generated data.
1 code implementation • IEEE Transactions on Pattern Analysis and Machine Intelligence 2021 • Wen-Yan Lin, Siying Liu, Changhao Ren, Ngai-Man Cheung, Hongdong Li, Yasuyuki Matsushita
The foundational assumption of machine learning is that the data under consideration is separable into classes; while intuitively reasonable, separability constraints have proven remarkably difficult to formulate mathematically.
Ranked #1 on Unsupervised Anomaly Detection with Specified Settings -- 10% anomaly on STL-10 (using extra training data)
1 code implementation • CVPR 2021 • Keshigeyan Chandrasegaran, Ngoc-Trung Tran, Ngai-Man Cheung
Our results prompt re-thinking of using high frequency Fourier spectrum decay attributes for CNN-generated image detection.
no code implementations • 1 Jan 2021 • Thilini Cooray, Ngai-Man Cheung, Wei Lu
Our work is the first to study disentanglement learning for graph-level representations.
no code implementations • 26 Dec 2020 • Tuan Hoang, Thanh-Toan Do, Tam V. Nguyen, Ngai-Man Cheung
With this approach, we can learn activation quantizers that minimize the quantization errors in the majority of channels.
no code implementations • 9 Dec 2020 • Yi Xiang Marcus Tan, Penny Chong, Jiamei Sun, Ngai-Man Cheung, Yuval Elovici, Alexander Binder
In this work, we propose a detection strategy to identify adversarial support sets, aimed at destroying the understanding of a few-shot classifier for a certain class.
no code implementations • 11 Nov 2020 • Penny Chong, Ngai-Man Cheung, Yuval Elovici, Alexander Binder
We compare performances in terms of the classification, explanation quality, and outlier detection of our proposed network with other baselines.
no code implementations • 1 Aug 2020 • Tuan Hoang, Thanh-Toan Do, Tam V. Nguyen, Ngai-Man Cheung
This paper presents a novel framework, namely Deep Cross-modality Spectral Hashing (DCSH), to tackle the unsupervised learning problem of binary hash codes for efficient cross-modal retrieval.
1 code implementation • 17 Jul 2020 • Jiamei Sun, Sebastian Lapuschkin, Wojciech Samek, Yunqing Zhao, Ngai-Man Cheung, Alexander Binder
It leverages on the explanation scores, obtained from existing explanation methods when applied to the predictions of FSC models, computed for intermediate feature maps of the models.
Ranked #8 on Cross-Domain Few-Shot on ISIC2018
1 code implementation • 9 Jul 2020 • Kwot Sin Lee, Ngoc-Trung Tran, Ngai-Man Cheung
While Generative Adversarial Networks (GANs) are fundamental to many generative modelling applications, they suffer from numerous issues.
1 code implementation • 9 Jun 2020 • Ngoc-Trung Tran, Viet-Hung Tran, Ngoc-Bao Nguyen, Trung-Kien Nguyen, Ngai-Man Cheung
We provide theoretical analysis to show that using our proposed DAG aligns with the original GAN in minimizing the Jensen-Shannon (JS) divergence between the original distribution and model distribution.
1 code implementation • NeurIPS 2019 • Ngoc-Trung Tran, Viet-Hung Tran, Ngoc-Bao Nguyen, Linxiao Yang, Ngai-Man Cheung
From the analysis, we identify issues of SS tasks which allow a severely mode-collapsed generator to excel the SS tasks.
Ranked #4 on Image Generation on ImageNet 32x32
no code implementations • 25 Sep 2019 • Yi Loo, Yiluan Guo, Ngai-Man Cheung
Recent few-shot learning algorithms have enabled models to quickly adapt to new tasks based on only a few training samples.
no code implementations • 6 Jul 2019 • Saeid Hosseini, Saeed Najafipour, Ngai-Man Cheung, Hongzhi Yin, Mohammad Reza Kangavari, Xiaofang Zhou
We can use the temporal and textual data of the nodes to compute the edge weights and then generate subgraphs with highly relevant nodes.
no code implementations • 14 May 2019 • Ngoc-Trung Tran, Viet-Hung Tran, Ngoc-Bao Nguyen, Ngai-Man Cheung
Importantly, we find out that simultaneously training the discriminator to classify the fake class from the pseudo-classes of real samples for the classification task will improve the discriminator and subsequently lead better guides to train generator.
no code implementations • 24 Apr 2019 • Thanh-Toan Do, Khoa Le, Tuan Hoang, Huu Le, Tam V. Nguyen, Ngai-Man Cheung
This global vector is then subjected to a hashing function to generate a binary hash code.
1 code implementation • 6 Apr 2019 • Huu Le, Thanh-Toan Do, Tuan Hoang, Ngai-Man Cheung
In particular, our work enables the use of randomized methods for point cloud registration without the need of putative correspondences.
no code implementations • ICLR Workshop LLD 2019 • Yi Loo, Swee Kiat Lim, Gemma Roig, Ngai-Man Cheung
We show that our model outperforms the current state of the art meta-learning methods in various regression tasks.
no code implementations • 1 Mar 2019 • Mostafa Rezazad, Matthias R. Brust, Mohammad Akbari, Pascal Bouvry, Ngai-Man Cheung
A novel class of extreme link-flooding DDoS (Distributed Denial of Service) attacks is designed to cut off entire geographical areas such as cities and even countries from the Internet by simultaneously targeting a selected set of network links.
1 code implementation • 4 Nov 2018 • Ngoc-Trung Tran, Tuan-Anh Bui, Ngai-Man Cheung
Second, we propose a new technique, gradient matching, to align the distributions of the generated samples and the real samples.
no code implementations • 23 Aug 2018 • Swee Kiat Lim, Yi Loo, Ngoc-Trung Tran, Ngai-Man Cheung, Gemma Roig, Yuval Elovici
To the best of our knowledge, our method is the first data augmentation technique focused on improving performance in unsupervised anomaly detection.
no code implementations • 22 Aug 2018 • Sibo Song, Ngai-Man Cheung, Vijay Chandrasekhar, Bappaditya Mandal
Specifically, using frame-level features, DATP regresses importance of different temporal segments and generates weights for them.
no code implementations • 6 Aug 2018 • Sibo Song, Yueru Chen, Ngai-Man Cheung, C. -C. Jay Kuo
Therefore, we propose a Saak transform based preprocessing method with three steps: 1) transforming an input image to a joint spatial-spectral representation via the forward Saak transform, 2) apply filtering to its high-frequency components, and, 3) reconstructing the image via the inverse Saak transform.
no code implementations • CVPR 2018 • Yiluan Guo, Ngai-Man Cheung
In this work, we propose an efficient, end-to-end fully convolutional Siamese network that computes the similarities at multiple levels.
1 code implementation • ECCV 2018 • Ngoc-Trung Tran, Tuan-Anh Bui, Ngai-Man Cheung
We use this constraint to explicitly prevent the generator from mode collapse.
Ranked #18 on Image Generation on STL-10
no code implementations • 28 Feb 2018 • Hossein Nejati, Hamed Alizadeh Ghazijahani, Milad Abdollahzadeh, Tooba Malekzadeh, Ngai-Man Cheung, Kheng Hock Lee, Lian Leng Low
In particular, seemingly, all previous approaches have assumed only 3 tissue types in the chronic wounds, while these wounds commonly exhibit 7 distinct tissue types that presence of each one changes the treatment procedure.
no code implementations • 21 Feb 2018 • Thanh-Toan Do, Tuan Hoang, Dang-Khoa Le Tan, Trung Pham, Huu Le, Ngai-Man Cheung, Ian Reid
However, training deep hashing networks for the task is challenging due to the binary constraints on the hash codes, the similarity preserving property, and the requirement for a vast amount of labelled images.
no code implementations • 19 Feb 2018 • Tuan Hoang, Thanh-Toan Do, Huu Le, Dang-Khoa Le-Tan, Ngai-Man Cheung
For unsupervised data-dependent hashing, the two most important requirements are to preserve similarity in the low-dimensional feature space and to minimize the binary quantization loss.
no code implementations • 10 Feb 2018 • Ngoc-Trung Tran, Dang-Khoa Le Tan, Anh-Dzung Doan, Thanh-Toan Do, Tuan-Anh Bui, Mengxuan Tan, Ngai-Man Cheung
In order to overcome the resource constraints of mobile devices, we propose a system design that leverages the scalability advantage of image retrieval and accuracy of 3D model-based localization.
1 code implementation • 7 Feb 2018 • Thanh-Toan Do, Tuan Hoang, Dang-Khoa Le Tan, Huu Le, Tam V. Nguyen, Ngai-Man Cheung
In the large-scale image retrieval task, the two most important requirements are the discriminability of image representations and the efficiency in computation and storage of representations.
no code implementations • 31 Dec 2017 • Saurabh Misra, Mengxuan Tan, Mostafa Rezazad, Matthias R. Brust, Ngai-Man Cheung
The adoption of benign traffic, while simultaneously targeting multiple network links, makes the detection of the Crossfire attack a serious challenge.
Cryptography and Security
no code implementations • 26 Dec 2017 • Touba Malekzadeh, Milad Abdollahzadeh, Hossein Nejati, Ngai-Man Cheung
To ensure flight safety of aircraft structures, it is necessary to have regular maintenance using visual and nondestructive inspection (NDI) methods.
no code implementations • 8 Dec 2017 • Thanh-Toan Do, Tuan Hoang, Dang-Khoa Le Tan, Anh-Dzung Doan, Ngai-Man Cheung
This design has overcome a challenging problem in some previous works: optimizing non-smooth objective functions because of binarization.
no code implementations • 4 Dec 2017 • Yiren Zhou, Seyed-Mohsen Moosavi-Dezfooli, Ngai-Man Cheung, Pascal Frossard
First, we propose a measurement to estimate the effect of parameter quantization errors in individual layers on the overall model prediction accuracy.
no code implementations • 24 Nov 2017 • Dang-Khoa Le Tan, Thanh-Toan Do, Ngai-Man Cheung
Image hashing is a popular technique applied to large scale content-based visual retrieval due to its compact and efficient binary codes.
no code implementations • ICCV 2017 • Xin Sun, Ngai-Man Cheung, Hongxun Yao, Yiluan Guo
Part-based trackers are effective in exploiting local details of the target object for robust tracking.
1 code implementation • 4 Jul 2017 • Tuan Hoang, Thanh-Toan Do, Dang-Khoa Le Tan, Ngai-Man Cheung
Recent work adopts fine-tuned strategies to further improve the discriminative power of the descriptors.
no code implementations • 13 May 2017 • Yiluan Guo, Hossein Nejati, Ngai-Man Cheung
In particular, our work proposes a new deep neural network design that integrates graph information such as brain connectivity with fully-connected layers.
no code implementations • 6 Apr 2017 • Tuan Hoang, Thanh-Toan Do, Dang-Khoa Le Tan, Ngai-Man Cheung
We introduce a novel approach to improve unsupervised hashing.
no code implementations • CVPR 2017 • Thanh-Toan Do, Dang-Khoa Le Tan, Trung T. Pham, Ngai-Man Cheung
This feature vector is then subjected to a hashing function that produces a binary hash code.
no code implementations • 8 Jan 2017 • Yiren Zhou, Sibo Song, Ngai-Man Cheung
Image blur and image noise are common distortions during image acquisition.
no code implementations • 26 Sep 2016 • Rui Liu, Hossein Nejati, Seyed Hamid Safavi, Ngai-Man Cheung
We propose an algorithm to uncover the intrinsic low-rank component of a high-dimensional, graph-smooth and grossly-corrupted dataset, under the situations that the underlying graph is unknown.
no code implementations • 19 Jul 2016 • Thanh-Toan Do, Anh-Dzung Doan, Duc-Thanh Nguyen, Ngai-Man Cheung
This paper proposes two approaches for inferencing binary codes in two-step (supervised, unsupervised) hashing.
no code implementations • 18 Jul 2016 • Thanh-Toan Do, Anh-Dzung Doan, Ngai-Man Cheung
Our resulting optimization with these binary, independence, and balance constraints is difficult to solve.
no code implementations • 23 May 2016 • Thanh-Toan Do, Ngai-Man Cheung
The objective of this paper is to design an embedding method that maps local features describing an image (e. g. SIFT) to a higher dimensional representation useful for the image retrieval problem.
no code implementations • 25 Jan 2016 • Sibo Song, Ngai-Man Cheung, Vijay Chandrasekhar, Bappaditya Mandal, Jie Lin
With the increasing availability of wearable devices, research on egocentric activity recognition has received much attention recently.
no code implementations • 6 Jan 2016 • Yiren Zhou, Hossein Nejati, Thanh-Toan Do, Ngai-Man Cheung, Lynette Cheah
We address the vehicle detection and classification problems using Deep Neural Networks (DNNs) approaches.
no code implementations • 28 Aug 2015 • Thanh-Toan Do, Anh-Zung Doan, Ngai-Man Cheung
This paper addresses the problem of learning binary hash codes for large scale image search by proposing a novel hashing method based on deep neural network.
no code implementations • CVPR 2015 • Thanh-Toan Do, Quang D. Tran, Ngai-Man Cheung
The embedded vectors resulted by the function approximation process are then aggregated to form a single representation used in the image retrieval framework.
no code implementations • 4 Nov 2013 • Martin Takáč, Selin Damla Ahipaşaoğlu, Ngai-Man Cheung, Peter Richtárik
Our approach attacks the maximization problem in sparse PCA directly and is scalable to high-dimensional data.