Hwee Kuan Lee

Zerui Chen, Sonia Xhyn Teo, Andrie Ochtman et al.

A key technology enabling the success of catheter ablation treatment for atrial tachycardia is activation mapping, which relies on manual local activation time (LAT) annotation of all acquired intracardiac electrogram (EGM) signals. This is a time-consuming and error-prone procedure, due to the difficulty in identifying the signal activation peaks for fractionated signals. This work presents a Deep Learning approach for the automated classification of EGM signals into three different types: normal, abnormal, and unclassified, which forms part of the LAT annotation pipeline, and contributes towards bypassing the need for manual annotations of the LAT. The Deep Learning network, the CNN-LSTM model, is a hybrid network architecture which combines convolutional neural network (CNN) layers with long short-term memory (LSTM) layers. 1452 EGM signals from a total of 9 patients undergoing clinically-indicated 3D cardiac mapping were used for the training, validation and testing of our models. From our findings, the CNN-LSTM model achieved an accuracy of 81% for the balanced dataset. For comparison, we separately developed a rule-based Decision Trees model which attained an accuracy of 67% for the same balanced dataset. Our work elucidates that analysing the EGM signals using a set of explicitly specified rules as proposed by the Decision Trees model is not suitable as EGM signals are complex. The CNN-LSTM model, on the other hand, has the ability to learn the complex, intrinsic features within the signals and identify useful features to differentiate the EGM signals.

Mahsa Paknezhad, Hamsawardhini Rengarajan, Chenghao Yuan et al.

This paper takes a parallel learning approach for robust and transparent AI. A deep neural network is trained in parallel on multiple tasks, where each task is trained only on a subset of the network resources. Each subset consists of network segments, that can be combined and shared across specific tasks. Tasks can share resources with other tasks, while having independent task-related network resources. Therefore, the trained network can share similar representations across various tasks, while also enabling independent task-related representations. The above allows for some crucial outcomes. (1) The parallel nature of our approach negates the issue of catastrophic forgetting. (2) The sharing of segments uses network resources more efficiently. (3) We show that the network does indeed use learned knowledge from some tasks in other tasks, through shared representations. (4) Through examination of individual task-related and shared representations, the model offers transparency in the network and in the relationships across tasks in a multi-task setting. Evaluation of the proposed approach against complex competing approaches such as Continual Learning, Neural Architecture Search, and Multi-task learning shows that it is capable of learning robust representations. This is the first effort to train a DL model on multiple tasks in parallel. Our code is available at https://github.com/MahsaPaknezhad/PaRT

Mahsa Paknezhad, Cuong Phuc Ngo, Amadeus Aristo Winarto et al.

Despite many proposed algorithms to provide robustness to deep learning (DL) models, DL models remain susceptible to adversarial attacks. We hypothesize that the adversarial vulnerability of DL models stems from two factors. The first factor is data sparsity which is that in the high dimensional input data space, there exist large regions outside the support of the data distribution. The second factor is the existence of many redundant parameters in the DL models. Owing to these factors, different models are able to come up with different decision boundaries with comparably high prediction accuracy. The appearance of the decision boundaries in the space outside the support of the data distribution does not affect the prediction accuracy of the model. However, it makes an important difference in the adversarial robustness of the model. We hypothesize that the ideal decision boundary is as far as possible from the support of the data distribution. In this paper, we develop a training framework to observe if DL models are able to learn such a decision boundary spanning the space around the class distributions further from the data points themselves. Semi-supervised learning was deployed during training by leveraging unlabeled data generated in the space outside the support of the data distribution. We measured adversarial robustness of the models trained using this training framework against well-known adversarial attacks and by using robustness metrics. We found that models trained using our framework, as well as other regularization methods and adversarial training support our hypothesis of data sparsity and that models trained with these methods learn to have decision boundaries more similar to the aforementioned ideal decision boundary. The code for our training framework is available at https://github.com/MahsaPaknezhad/AdversariallyRobustTraining.

Mahsa Paknezhad, Sheng Yang Michael Loh, Yukti Choudhury et al.

Motivation: High resolution 2D whole slide imaging provides rich information about the tissue structure. This information can be a lot richer if these 2D images can be stacked into a 3D tissue volume. A 3D analysis, however, requires accurate reconstruction of the tissue volume from the 2D image stack. This task is not trivial due to the distortions that each individual tissue slice experiences while cutting and mounting the tissue on the glass slide. Performing registration for the whole tissue slices may be adversely affected by the deformed tissue regions. Consequently, regional registration is found to be more effective. In this paper, we propose an accurate and robust regional registration algorithm for whole slide images which incrementally focuses registration on the area around the region of interest. Results: Using mean similarity index as the metric, the proposed algorithm (mean $\pm$ std: $0.84 \pm 0.11$) followed by a fine registration algorithm ($0.86 \pm 0.08$) outperformed the state-of-the-art linear whole tissue registration algorithm ($0.74 \pm 0.19$) and the regional version of this algorithm ($0.81 \pm 0.15$). The proposed algorithm also outperforms the state-of-the-art nonlinear registration algorithm (original : $0.82 \pm 0.12$, regional : $0.77 \pm 0.22$) for whole slide images and a recently proposed patch-based registration algorithm (patch size 256: $0.79 \pm 0.16$ , patch size 512: $0.77 \pm 0.16$) for medical images. Availability: The C++ implementation code is available online at the github repository: https://github.com/MahsaPaknezhad/WSIRegistration

Kerem Delikoyun, Qianyu Chen, Liu Wei et al.

While analysing rare blood cell aggregates remains challenging in automated haematology, they could markedly advance label-free functional diagnostics. Conventional flow cytometers efficiently perform cell counting with leukocyte differentials but fail to identify aggregates with flagged results, requiring manual reviews. Quantitative phase imaging flow cytometry captures detailed aggregate morphologies, but clinical use is hampered by massive data storage and offline processing. Incorporating hidden biomarkers into routine haematology panels would significantly improve diagnostics without flagged results. We present RT-HAD, an end-to-end deep learning-based image and data processing framework for off-axis digital holographic microscopy (DHM), which combines physics-consistent holographic reconstruction and detection, representing each blood cell in a graph to recognize aggregates. RT-HAD processes >30 GB of image data on-the-fly with turnaround time of <1.5 min and error rate of 8.9% in platelet aggregate detection, which matches acceptable laboratory error rates of haematology biomarkers and solves the big data challenge for point-of-care diagnostics.

Haonan Wang, Brian Chen, Siquan Li et al.

Parameter-Efficient Fine-Tuning (PEFT) methods have become crucial for rapidly adapting large language models (LLMs) to downstream tasks. Prefix-Tuning, an early and effective PEFT technique, demonstrated the ability to achieve performance comparable to full fine-tuning with significantly reduced computational and memory overhead. However, despite its earlier success, its effectiveness in training modern state-of-the-art LLMs has been very limited. In this work, we demonstrate empirically that Prefix-Tuning underperforms on LLMs because of an inherent tradeoff between input and prefix significance within the attention head. This motivates us to introduce Prefix-Tuning+, a novel architecture that generalizes the principles of Prefix-Tuning while addressing its shortcomings by shifting the prefix module out of the attention head itself. We further provide an overview of our construction process to guide future users when constructing their own context-based methods. Our experiments show that, across a diverse set of benchmarks, Prefix-Tuning+ consistently outperforms existing Prefix-Tuning methods. Notably, it achieves performance on par with the widely adopted LoRA method on several general benchmarks, highlighting the potential modern extension of Prefix-Tuning approaches. Our findings suggest that by overcoming its inherent limitations, Prefix-Tuning can remain a competitive and relevant research direction in the landscape of parameter-efficient LLM adaptation.

Meenakshi Gupta, Mingyuan Lei, Tat-Jen Cham et al.

This paper introduces a novel framework named D-LORD (Double Latent Optimization for Representation Disentanglement), which is designed for motion stylization (motion style transfer and motion retargeting). The primary objective of this framework is to separate the class and content information from a given motion sequence using a data-driven latent optimization approach. Here, class refers to person-specific style, such as a particular emotion or an individual's identity, while content relates to the style-agnostic aspect of an action, such as walking or jumping, as universally understood concepts. The key advantage of D-LORD is its ability to perform style transfer without needing paired motion data. Instead, it utilizes class and content labels during the latent optimization process. By disentangling the representation, the framework enables the transformation of one motion sequences style to another's style using Adaptive Instance Normalization. The proposed D-LORD framework is designed with a focus on generalization, allowing it to handle different class and content labels for various applications. Additionally, it can generate diverse motion sequences when specific class and content labels are provided. The framework's efficacy is demonstrated through experimentation on three datasets: the CMU XIA dataset for motion style transfer, the MHAD dataset, and the RRIS Ability dataset for motion retargeting. Notably, this paper presents the first generalized framework for motion style transfer and motion retargeting, showcasing its potential contributions in this area.

Brian K Chen, Tianyang Hu, Hui Jin et al.

In-Context Learning (ICL) has been a powerful emergent property of large language models that has attracted increasing attention in recent years. In contrast to regular gradient-based learning, ICL is highly interpretable and does not require parameter updates. In this paper, we show that, for linearized transformer networks, ICL can be made explicit and permanent through the inclusion of bias terms. We mathematically demonstrate the equivalence between a model with ICL demonstration prompts and the same model with the additional bias terms. Our algorithm (ICLCA) allows for exact conversion in an inexpensive manner. Existing methods are not exact and require expensive parameter updates. We demonstrate the efficacy of our approach through experiments that show the exact incorporation of ICL tokens into a linear transformer. We further suggest how our method can be adapted to achieve cheap approximate conversion of ICL tokens, even in regular transformer networks that are not linearized. Our experiments on GPT-2 show that, even though the conversion is only approximate, the model still gains valuable context from the included bias terms.

Jiyi Zhang, Han Fang, Hwee Kuan Lee et al.

Given a poorly documented neural network model, we take the perspective of a forensic investigator who wants to find out the model's data domain (e.g. whether on face images or traffic signs). Although existing methods such as membership inference and model inversion can be used to uncover some information about an unknown model, they still require knowledge of the data domain to start with. In this paper, we propose solving this problem by leveraging on comprehensive corpus such as ImageNet to select a meaningful distribution that is close to the original training distribution and leads to high performance in follow-up investigations. The corpus comprises two components, a large dataset of samples and meta information such as hierarchical structure and textual information on the samples. Our goal is to select a set of samples from the corpus for the given model. The core of our method is an objective function that considers two criteria on the selected samples: the model functional properties (derived from the dataset), and semantics (derived from the metadata). We also give an algorithm to efficiently search the large space of all possible subsets w.r.t. the objective function. Experimentation results show that the proposed method is effective. For example, cloning a given model (originally trained with CIFAR-10) by using Caltech 101 can achieve 45.5% accuracy. By using datasets selected by our method, the accuracy is improved to 72.0%.

Suvidha Tripathi, Satish Kumar Singh, Hwee Kuan Lee

Researchers working on computational analysis of Whole Slide Images (WSIs) in histopathology have primarily resorted to patch-based modelling due to large resolution of each WSI. The large resolution makes WSIs infeasible to be fed directly into the machine learning models due to computational constraints. However, due to patch-based analysis, most of the current methods fail to exploit the underlying spatial relationship among the patches. In our work, we have tried to integrate this relationship along with feature-based correlation among the extracted patches from the particular tumorous region. For the given task of classification, we have used BiLSTMs to model both forward and backward contextual relationship. RNN based models eliminate the limitation of sequence size by allowing the modelling of variable size images within a deep learning model. We have also incorporated the effect of spatial continuity by exploring different scanning techniques used to sample patches. To establish the efficiency of our approach, we trained and tested our model on two datasets, microscopy images and WSI tumour regions. After comparing with contemporary literature we achieved the better performance with accuracy of 90% for microscopy image dataset. For WSI tumour region dataset, we compared the classification results with deep learning networks such as ResNet, DenseNet, and InceptionV3 using maximum voting technique. We achieved the highest performance accuracy of 84%. We found out that BiLSTMs with CNN features have performed much better in modelling patches into an end-to-end Image classification network. Additionally, the variable dimensions of WSI tumour regions were used for classification without the need for resizing. This suggests that our method is independent of tumour image size and can process large dimensional images without losing the resolution details.

Chengyang Zhou, Thao Vy Dinh, Heyi Kong et al.

The evaluation of obstructions (stenosis) in coronary arteries is currently done by a physician's visual assessment of coronary angiography video sequences. It is laborious, and can be susceptible to interobserver variation. Prior studies have attempted to automate this process, but few have demonstrated an integrated suite of algorithms for the end-to-end analysis of angiograms. We report an automated analysis pipeline based on deep learning to rapidly and objectively assess coronary angiograms, highlight coronary vessels of interest, and quantify potential stenosis. We propose a 3-stage automated analysis method consisting of key frame extraction, vessel segmentation, and stenosis measurement. We combined powerful deep learning approaches such as ResNet and U-Net with traditional image processing and geometrical analysis. We trained and tested our algorithms on the Left Anterior Oblique (LAO) view of the right coronary artery (RCA) using anonymized angiograms obtained from a tertiary cardiac institution, then tested the generalizability of our technique to the Right Anterior Oblique (RAO) view. We demonstrated an overall improvement on previous work, with key frame extraction top-5 precision of 98.4%, vessel segmentation F1-Score of 0.891 and stenosis measurement 20.7% Type I Error rate.

Yusuke Tomita, Kenta Shiina, Yutaka Okabe et al.

We use the Fortuin-Kasteleyn representation based improved estimator of the correlation configuration as an alternative to the ordinary correlation configuration in the machine-learning study of the phase classification of spin models. The phases of classical spin models are classified using the improved estimators, and the method is also applied to the quantum Monte Carlo simulation using the loop algorithm. We analyze the Berezinskii-Kosterlitz-Thouless (BKT) transition of the spin 1/2 quantum XY model on the square lattice. We classify the BKT phase and the paramagnetic phase of the quantum XY model using the machine-learning approach. We show that the classification of the quantum XY model can be performed by using the training data of the classical XY model.

Mustafa Umit Oner, Jared Marc Song Kye-Jet, Hwee Kuan Lee et al.

There are different multiple instance learning (MIL) pooling filters used in MIL models. In this paper, we study the effect of different MIL pooling filters on the performance of MIL models in real world MIL tasks. We designed a neural network based MIL framework with 5 different MIL pooling filters: `max', `mean', `attention', `distribution' and `distribution with attention'. We also formulated 5 different MIL tasks on a real world lymph node metastases dataset. We found that the performance of our framework in a task is different for different filters. We also observed that the performances of the five pooling filters are also different from task to task. Hence, the selection of a correct MIL pooling filter for each MIL task is crucial for better performance. Furthermore, we noticed that models with `distribution' and `distribution with attention' pooling filters consistently perform well in almost all of the tasks. We attribute this phenomena to the amount of information captured by `distribution' based pooling filters. While point estimate based pooling filters, like `max' and `mean', produce point estimates of distributions, `distribution' based pooling filters capture the full information in distributions. Lastly, we compared the performance of our neural network model with `distribution' pooling filter with the performance of the best MIL methods in the literature on classical MIL datasets and our model outperformed the others.

Jiyi Zhang, Ee-Chien Chang, Hwee Kuan Lee

Many machine learning adversarial attacks find adversarial samples of a victim model ${\mathcal M}$ by following the gradient of some attack objective functions, either explicitly or implicitly. To confuse and detect such attacks, we take the proactive approach that modifies those functions with the goal of misleading the attacks to some local minimals, or to some designated regions that can be easily picked up by an analyzer. To achieve this goal, we propose adding a large number of artifacts, which we called $attractors$, onto the otherwise smooth function. An attractor is a point in the input space, where samples in its neighborhood have gradient pointing toward it. We observe that decoders of watermarking schemes exhibit properties of attractors and give a generic method that injects attractors from a watermark decoder into the victim model ${\mathcal M}$. This principled approach allows us to leverage on known watermarking schemes for scalability and robustness and provides explainability of the outcomes. Experimental studies show that our method has competitive performance. For instance, for un-targeted attacks on CIFAR-10 dataset, we can reduce the overall attack success rate of DeepFool to 1.9%, whereas known defense LID, FS and MagNet can reduce the rate to 90.8%, 98.5% and 78.5% respectively.

Malay Singh, Emarene Mationg Kalaw, Wang Jie et al.

Architecture, size, and shape of glands are most important patterns used by pathologists for assessment of cancer malignancy in prostate histopathological tissue slides. Varying structures of glands along with cumbersome manual observations may result in subjective and inconsistent assessment. Cribriform gland with irregular border is an important feature in Gleason pattern 4. We propose using deep neural networks for cribriform pattern classification in prostate histopathological images. $163708$ Hematoxylin and Eosin (H\&E) stained images were extracted from histopathologic tissue slides of $19$ patients with prostate cancer and annotated for cribriform patterns. Our automated image classification system analyses the H\&E images to classify them as either `Cribriform' or `Non-cribriform'. Our system uses various deep learning approaches and hand-crafted image pixel intensity-based features. We present our results for cribriform pattern detection across various parameters and configuration allowed by our system. The combination of fine-tuned deep learning models outperformed the state-of-art nuclei feature based methods. Our image classification system achieved the testing accuracy of $85.93~\pm~7.54$ (cross-validated) and $88.04~\pm~5.63$ ( additional unseen test set) across three folds. In this paper, we present an annotated cribriform dataset along with analysis of deep learning models and hand-crafted features for cribriform pattern detection in prostate histopathological images.

Mustafa Umit Oner, Hwee Kuan Lee, Wing-Kin Sung

A weakly supervised learning based clustering framework is proposed in this paper. As the core of this framework, we introduce a novel multiple instance learning task based on a bag level label called unique class count ($ucc$), which is the number of unique classes among all instances inside the bag. In this task, no annotations on individual instances inside the bag are needed during training of the models. We mathematically prove that with a perfect $ucc$ classifier, perfect clustering of individual instances inside the bags is possible even when no annotations on individual instances are given during training. We have constructed a neural network based $ucc$ classifier and experimentally shown that the clustering performance of our framework with our weakly supervised $ucc$ classifier is comparable to that of fully supervised learning models where labels for all instances are known. Furthermore, we have tested the applicability of our framework to a real world task of semantic segmentation of breast cancer metastases in histological lymph node sections and shown that the performance of our weakly supervised framework is comparable to the performance of a fully supervised Unet model.

Connie Kou, Hwee Kuan Lee, Ee-Chien Chang et al.

Adversarial attacks on convolutional neural networks (CNN) have gained significant attention and there have been active research efforts on defense mechanisms. Stochastic input transformation methods have been proposed, where the idea is to recover the image from adversarial attack by random transformation, and to take the majority vote as consensus among the random samples. However, the transformation improves the accuracy on adversarial images at the expense of the accuracy on clean images. While it is intuitive that the accuracy on clean images would deteriorate, the exact mechanism in which how this occurs is unclear. In this paper, we study the distribution of softmax induced by stochastic transformations. We observe that with random transformations on the clean images, although the mass of the softmax distribution could shift to the wrong class, the resulting distribution of softmax could be used to correct the prediction. Furthermore, on the adversarial counterparts, with the image transformation, the resulting shapes of the distribution of softmax are similar to the distributions from the clean images. With these observations, we propose a method to improve existing transformation-based defenses. We train a separate lightweight distribution classifier to recognize distinct features in the distributions of softmax outputs of transformed images. Our empirical studies show that our distribution classifier, by training on distributions obtained from clean images only, outperforms majority voting for both clean and adversarial images. Our method is generic and can be integrated with existing transformation-based defenses.

Cuong Phuc Ngo, Amadeus Aristo Winarto, Connie Kou Khor Li et al.

Anomaly detection is a classical problem where the aim is to detect anomalous data that do not belong to the normal data distribution. Current state-of-the-art methods for anomaly detection on complex high-dimensional data are based on the generative adversarial network (GAN). However, the traditional GAN loss is not directly aligned with the anomaly detection objective: it encourages the distribution of the generated samples to overlap with the real data and so the resulting discriminator has been found to be ineffective as an anomaly detector. In this paper, we propose simple modifications to the GAN loss such that the generated samples lie at the boundary of the real data distribution. With our modified GAN loss, our anomaly detection method, called Fence GAN (FGAN), directly uses the discriminator score as an anomaly threshold. Our experimental results using the MNIST, CIFAR10 and KDD99 datasets show that Fence GAN yields the best anomaly classification accuracy compared to state-of-the-art methods.

Mundher Al-Shabi, Hwee Kuan Lee, Maxine Tan

Different types of Convolutional Neural Networks (CNNs) have been applied to detect cancerous lung nodules from computed tomography (CT) scans. However, the size of a nodule is very diverse and can range anywhere between 3 and 30 millimeters. The high variation of nodule sizes makes classifying them a difficult and challenging task. In this study, we propose a novel CNN architecture called Gated-Dilated (GD) networks to classify nodules as malignant or benign. Unlike previous studies, the GD network uses multiple dilated convolutions instead of max-poolings to capture the scale variations. Moreover, the GD network has a Context-Aware sub-network that analyzes the input features and guides the features to a suitable dilated convolution. We evaluated the proposed network on more than 1,000 CT scans from the LIDC-LDRI dataset. Our proposed network outperforms state-of-the-art baseline models including Multi-Crop, Resnet, and Densenet, with an AUC of >0.95. Compared to the baseline models, the GD network improves the classification accuracies of mid-range sized nodules. Furthermore, we observe a relationship between the size of the nodule and the attention signal generated by the Context-Aware sub-network, which validates our new network architecture.

Connie Kou, Hwee Kuan Lee, Jorge Sanz et al.

There is emerging interest in performing regression between distributions. In contrast to prediction on single instances, these machine learning methods can be useful for population-based studies or on problems that are inherently statistical in nature. The recently proposed distribution regression network (DRN) has shown superior performance for the distribution-to-distribution regression task compared to conventional neural networks. However, in Kou et al. (2018) and some other works on distribution regression, there is a lack of comprehensive comparative study on both theoretical basis and generalization abilities of the methods. We derive some mathematical properties of DRN and qualitatively compare it to conventional neural networks. We also perform comprehensive experiments to study the generalizability of distribution regression models, by studying their robustness to limited training data, data sampling noise and task difficulty. DRN consistently outperforms conventional neural networks, requiring fewer training data and maintaining robust performance with noise. Furthermore, the theoretical properties of DRN can be used to provide some explanation on the ability of DRN to achieve better generalization performance than conventional neural networks.

Connie Kou, Hwee Kuan Lee, Teck Khim Ng

Despite the superior performance of deep learning in many applications, challenges remain in the area of regression on function spaces. In particular, neural networks are unable to encode function inputs compactly as each node encodes just a real value. We propose a novel idea to address this shortcoming: to encode an entire function in a single network node. To that end, we design a compact network representation that encodes and propagates functions in single nodes for the distribution regression task. Our proposed Distribution Regression Network (DRN) achieves higher prediction accuracies while being much more compact and uses fewer parameters than traditional neural networks.

Jiyi Zhang, Hung Dang, Hwee Kuan Lee et al.

We propose a flipped-Adversarial AutoEncoder (FAAE) that simultaneously trains a generative model G that maps an arbitrary latent code distribution to a data distribution and an encoder E that embodies an "inverse mapping" that encodes a data sample into a latent code vector. Unlike previous hybrid approaches that leverage adversarial training criterion in constructing autoencoders, FAAE minimizes re-encoding errors in the latent space and exploits adversarial criterion in the data space. Experimental evaluations demonstrate that the proposed framework produces sharper reconstructed images while at the same time enabling inference that captures rich semantic representation of data.