| Total: 30
Immune repertoire classification, a typical multiple instance learning (MIL) problem, is a frontier research topic in computational biology that makes transformative contributions to new vaccines and immune therapies. However, the traditional instance-space MIL, directly assigning bag-level labels to instances, suffers from the massive amount of noisy labels and extremely low witness rate. In this work, we propose a noisy-label-learning formulation to solve the immune repertoire classification task. To remedy the inaccurate supervision of repertoire-level labels for a sequence-level classifier, we design a robust training strategy: The initial labels are smoothed to be asymmetric and are progressively corrected using the model's predictions throughout the training process. Furthermore, two models with the same architecture but different parameter initialization are co-trained simultaneously to remedy the known ``confirmation bias'' problem in the self-training-like schema. As a result, we obtain accurate sequence-level classification and, subsequently, repertoire-level classification. Experiments on the Cytomegalovirus (CMV) and Cancer datasets demonstrate our method's effectiveness and superior performance on sequence-level and repertoire-level tasks. Code available at https://github.com/TencentAILabHealthcare/NLL-IRC.
Machine-learning models are becoming increasingly prevalent in our lives, for instance assisting in image-classification or decision-making tasks. Consequently, the reliability of these models is of critical importance and has resulted in the development of numerous approaches for validating and verifying their robustness and fairness. However, beyond such specific properties, it is challenging to specify, let alone check, general functional-correctness expectations from models. In this paper, we take inspiration from specifications used in formal methods, expressing functional-correctness properties by reasoning about k different executions---so-called k-safety properties. Considering a credit-screening model of a bank, the expected property that "if a person is denied a loan and their income decreases, they should still be denied the loan" is a 2-safety property. Here, we show the wide applicability of k-safety properties for machine-learning models and present the first specification language for expressing them. We also operationalize the language in a framework for automatically validating such properties using metamorphic testing. Our experiments show that our framework is effective in identifying property violations, and that detected bugs could be used to train better models.
Recently, the wanton dissemination of fake news on social media has adversely affected our lives, rendering automatic fake news detection a pressing issue. Current methods are often fully supervised and typically employ deep neural networks (DNN) to learn implicit relevance from labeled data, ignoring explicitly shared properties (e.g., inflammatory expressions) across fake news. To address this limitation, we propose a graph-theoretic framework, called Generalized Deep Markov Random Fields Framework (GDMRFF), that inherits the capability of deep learning while at the same time exploiting the correlations among the news articles (including labeled and unlabeled data). Specifically, we first leverage a DNN-based module to learn implicit relations, which we then reveal as the unary function of MRF. Pairwise functions with refining effects to encapsulate human insights are designed to capture the explicit association among all samples. Meanwhile, an event removal module is introduced to remove event impact on pairwise functions. Note that we train GDMRFF with the semi-supervised setting, which decreases the reliance on labeled data while maximizing the potential of unlabeled data. We further develop an Ambiguity Learning Guided MRF (ALGM) model as a concretization of GDMRFF. Experiments show that ALGM outperforms the compared methods significantly on two datasets, especially when labeled data is limited.
Typical RL-for-finance solutions directly optimize trading policies over the noisy market data, such as stock prices and trading volumes, without explicitly considering the future trends and correlations of different investment assets as we humans do. In this paper, we present StockFormer, a hybrid trading machine that integrates the forward modeling capabilities of predictive coding with the advantages of RL agents in policy flexibility. The predictive coding part consists of three Transformer branches with modified structures, which respectively extract effective latent states of long-/short-term future dynamics and asset relations. The RL agent adaptively fuses these states and then executes an actor-critic algorithm in the unified state space. The entire model is jointly trained by propagating the critic's gradients back to the predictive coding module. StockFormer significantly outperforms existing approaches across three publicly available financial datasets in terms of portfolio returns and Sharpe ratios.
Various strategies for label-scarce object detection have been explored by the computer vision research community. These strategies mainly rely on assumptions that are specific to natural images and not directly applicable to the biological and biomedical vision domains. For example, most semi-supervised learning strategies rely on a small set of labeled data as a confident source of ground truth. In many biological vision applications, however, the ground truth is unknown and indirect information might be available in the form of noisy estimations or orthogonal evidence. In this work, we frame a crucial problem in spatial transcriptomics - decoding barcodes from In-Situ-Sequencing (ISS) images - as a semi-supervised object detection (SSOD) problem. Our proposed framework incorporates additional available sources of information into a semi-supervised learning framework in the form of privileged information. The privileged information is incorporated into the teacher's pseudo-labeling in a teacher-student self-training iteration. Although the available privileged information could be data domain specific, we have introduced a general strategy of pseudo-labeling enhanced by privileged information (PLePI) and exemplified the concept using ISS images, as well on the COCO benchmark using extra evidence provided by CLIP.
It is a common practice for designers to create digital prototypes from a mock-up/screenshot. Reverse engineering graphic design by detecting its components (e.g., text, icon, button) helps expedite this process. This paper first conducts a statistical analysis to emphasize the importance of relations in graphic layouts, which further motivates us to incorporate relation modeling into component detection. Built on the current state-of-the-art DETR (DEtection TRansformer), we introduce a learnable relation matrix to model class correlations. Specifically, the matrix will be added in the DETR decoder to update the query-to-query self-attention. Experiment results on three public datasets show that our approach achieves better performance than several strong baselines. We further visualize the learnt relation matrix and observe some reasonable patterns. Moreover, we show an application of component detection where we leverage the detection outputs as augmented training data for layout generation, which achieves promising results.
Snapshot observation based source localization has been widely studied due to its accessibility and low cost. However, the interaction of users in existing methods does not be addressed in time-varying infection scenarios. So these methods have a decreased accuracy in heterogeneous interaction scenarios. To solve this critical issue, this paper proposes a sequence-to-sequence based localization framework called Temporal-sequence based Graph Attention Source Identification (TGASI) based on an inductive learning idea. More specifically, the encoder focuses on generating multiple features by estimating the influence probability between two users, and the decoder distinguishes the importance of prediction sources in different timestamps by a designed temporal attention mechanism. It's worth mentioning that the inductive learning idea ensures that TGASI can detect the sources in new scenarios without knowing other prior knowledge, which proves the scalability of TGASI. Comprehensive experiments with the SOTA methods demonstrate the higher detection performance and scalability in different scenarios of TGASI.
Set Cover is a fundamental problem in combinatorial optimization which has been studied for many decades due to its various applications across multiple domains. In many of these domains, the input data consists of locations, relationships, and other sensitive information of individuals which may leaked due to the set cover output. Attempts have been made to design privacy-preserving algorithms to solve the Set Cover under privacy constraints. Under differential privacy, it has been proved that the Set Cover problem has strong impossibility results and no explicit forms of the output can be released to the public. In this work, we observe that these hardness results dissolve when we turn to the Partial Set Cover problem, where we only need to cover a ρ ∈ (0,1) fraction of the elements. We show that this relaxation enables us to avoid the impossibility results, and give the first algorithm which outputs an explicit form of set cover with non-trivial utility guarantees under differential privacy. Using our algorithm as a subroutine, we design a differentially private bicriteria algorithm to solve a recently proposed facility location problem for vaccine distribution which generalizes the k-supplier with outliers. Our analysis shows that relaxing the covering requirement to serve only a ρ ∈ (0,1) fraction of the population/universe also allows us to circumvent the inherent hardness of k-supplier and give the first non-trivial guarantees.
Adversarial example is a rising tool for voice privacy protection. By adding imperceptible noise to public audio, it prevents tampers from using zero-shot Voice Conversion (VC) to synthesize high quality speech with target speaker identity. However, many existing studies ignore the human perception characteristics of audio data, and it is challenging to generate strong and imperceptible adversarial audio. In this paper, we propose the Voice Guard defense method, which uses a novel method to advance the adversarial perturbation to the time domain to avoid the loss caused by cross-domain conversion. And the psychoacoustic model is introduced into the defense of VC for the first time, which greatly improves the disruption ability and concealment of adversarial audio. We also standardize the evaluation metrics of adversarial audio for the first time, combining multi-dimensional metrics to define the criteria for defense. We evaluate Voice Guard on several state-of-the-art zero-shot VC models. The experimental results show that our method can ensure the perceptual quality of adversarial audio while having a strong defense capability, and is far superior to previous works in terms of disruption ability and concealment.
Protein binding site prediction is an important prerequisite for the discovery of new drugs. Usually, natural 3D U-Net is adopted as the standard site prediction framework to do per-voxel binary mask classification. However, this scheme only performs feature extraction for single-scale samples, which may bring the loss of global or local information, resulting in incomplete, artifacted or even missed predictions. To tackle this issue, we propose a network called GLPocket, which is based on the Lmser (Least mean square error reconstruction) network and utilizes multi-scale representation to predict binding sites. Firstly, GLPocket uses Target Cropping Block (TCB) for targeted prediction. TCB selects the local interested feature from the global representations to perform concentrated prediction, and reduces the volume of feature maps to be calculated by 82% without adding additional parameters. It integrates global distribution information into local regions, making prediction more concentrated on decoding stage. Secondly, GLPocket establishes long-range relationship of patches within the local region with Transformer Block (TB), to enrich local context semantic information. Experiments show that GLPocket improves by 0.5%-4% on DCA Top-n prediction compared with previous state-of-the-art methods on four datasets. Our code has been released in https://github.com/CMACH508/GLPocket.
Identifying the potential associations among drugs, microbes and diseases is of great significance in exploring the pathogenesis and improving precision medicine. There are plenty of computational methods for pair-wise association prediction, such as drug-microbe and microbe-disease associations, but few methods focus on the higher-order triple-wise drug-microbe-disease (DMD) associations. Driven by the advancement of hypergraph neural networks (HGNNs), we expect them to fully capture high-order interaction patterns behind the hypergraph formulated by DMD associations and realize sound prediction performance. However, the confirmed DMD associations are insufficient due to the high cost of in vitro screening, which forms a sparse DMD hypergraph and thus brings in suboptimal generalization ability. To mitigate the limitation, we propose a Multi-view Contrastive Learning Hypergraph Neural Network, named MCHNN, for DMD association prediction. We design a novel multi-view contrastive learning on the DMD hypergraph as an auxiliary task, which guides the HGNN to learn more discriminative representations and enhances the generalization ability. Extensive computational experiments show that MCHNN achieves satisfactory performance in DMD association prediction and, more importantly, demonstrate the effectiveness of our devised multi-view contrastive learning on the sparse DMD hypergraph.
Synthesis-based steganography without embedding (SWE) methods transform secret messages to container images synthesised by generative networks, which eliminates distortions of container images and thus can fundamentally resist typical steganalysis tools. However, existing methods suffer from weak message recovery robustness, synthesis fidelity, and the risk of message leakage. To address these problems, we propose a novel robust steganography without embedding method in this paper. In particular, we design a secure weight modulation-based generator by introducing secure factors to hide secret messages in synthesised container images. In this manner, the synthesised results are modulated by secure factors and thus the secret messages are inaccessible when using fake factors, thus reducing the risk of message leakage. Furthermore, we design a difference predictor via the reconstruction of tampered container images together with an adversarial training strategy to iteratively update the estimation of hidden messages. This ensures robustness of recovering hidden messages, while degradation of synthesis fidelity is reduced since the generator is not included in the adversarial training. Extensive experimental results convincingly demonstrate that our proposed method is effective in avoiding message leakage and superior to other existing methods in terms of recovery robustness and synthesis fidelity.
This paper introduces Local Learner (2L), an algorithm for providing a set of reference strategies to guide the search for programmatic strategies in two-player zero-sum games. Previous learning algorithms, such as Iterated Best Response (IBR), Fictitious Play (FP), and Double-Oracle (DO), can be computationally expensive or miss important information for guiding search algorithms. 2L actively selects a set of reference strategies to improve the search signal. We empirically demonstrate the advantages of our approach while guiding a local search algorithm for synthesizing strategies in three games, including MicroRTS, a challenging real-time strategy game. Results show that 2L learns reference strategies that provide a stronger search signal than IBR, FP, and DO. We also simulate a tournament of MicroRTS, where a synthesizer using 2L outperformed the winners of the two latest MicroRTS competitions, which were programmatic strategies written by human programmers.
The deep clustering paradigm has shown great potential for discovering complex patterns that can reveal cell heterogeneity in single-cell RNA sequencing data. This paradigm involves two training phases: pretraining based on a pretext task and fine-tuning using pseudo-labels. Although current models yield promising results, they overlook the geometric distortions that regularly occur during the training process. More precisely, the transition between the two phases results in a coarse flattening of the latent structures, which can deteriorate the clustering performance. In this context, existing methods perform euclidean-based embedding clustering without ensuring the flatness and convexity of the latent manifolds. To address this problem, we incorporate two mechanisms. First, we introduce an overclustering loss to flatten the local curves. Second, we propose an adversarial mechanism to adjust the global geometric configuration. The second mechanism gradually transforms the latent structures into convex ones. Empirical results on a variety of gene expression datasets show that our model outperforms state-of-the-art methods.
In recent years, the state-of-the-art agents for playing abstract board games, like chess and others, have moved from using intricate hand-crafted models for evaluating the merits of individual game states toward using neural networks (NNs). This development has eased the encapsulation of the relevant domain-specific knowledge and resulted in much-improved playing strength. However, this has come at the cost of making the resulting models ill-interpretable and challenging to understand and use for enhancing human knowledge. Using a world-class superhuman-strength chess-playing engine as our testbed, we show how recent model probing interpretability techniques can shed light on concepts learned by the engine's NN. Furthermore, to gain additional insight, we contrast the game-state evaluations of the NN to that of its counterpart hand-crafted evaluation model and identify and explain some of the main differences.
A high-quality program generator is essential to effective automated compiler testing. Engineering such a program generator is difficult, time-consuming, and specific to the language under testing, thus requiring tremendous efforts from human experts with language-specific domain knowledge. To avoid repeatedly writing program generators for different languages, researchers recently proposed a language-agnostic approach based on deep learning techniques to automatically learn a program generator (referred to as DLG) from existing programs. Evaluations show that DLGs outperform Language-Specific Program Generators (LSGs) in testing compilers. However, we argue that it is unfair to use LSGs as baselines to evaluate DLGs. LSGs aim to validate compiler optimizations by only generating compilable, well-defined test programs; this restriction inevitably impairs the diversity of the language features used in the generated programs. In contrast, DLGs do not aim to validate the correctness of compiler optimizations, and its generated programs are not guaranteed to be well-defined or even compilable. Therefore, it is not surprising that DLG-generated programs are more diverse in terms of used language features than LSG-generated ones. This study revisits the evaluation of DLGs, and proposes a new, fair, simple yet strong baseline named Kitten for evaluating DLGs. Given a dataset consisting of human-written programs, instead of using deep learning techniques to learn a program generator, Kitten directly derives new programs by mutating the programs in the dataset. Extensive experiments with more than 1,500 CPU-hours demonstrate that the state-of-the-art DLGs fail to compete against such a simple baseline: 3 v.s. 1,750 hang bugs, 1 v.s. 34 distinct compiler crashes. We believe that DLGs still have a large room for improvement.
Advancements in reinforcement learning (RL) have demonstrated superhuman performance in complex tasks such as Starcraft, Go, Chess etc. However, knowledge transfer from Artificial ``Experts" to humans remain a significant challenge. A promising avenue for such transfer would be the use of curricula. Recent methods in curricula generation focuses on training RL agents efficiently, yet such methods rely on surrogate measures to track student progress, and are not suited for training robots in the real world (or more ambitiously humans). In this paper, we introduce a method named Parameterized Environment Response Model (PERM) that shows promising results in training RL agents in parameterized environments. Inspired by Item Response Theory, PERM seeks to model difficulty of environments and ability of RL agents directly. Given that RL agents and humans are trained more efficiently under the ``zone of proximal development", our method generates a curriculum by matching the difficulty of an environment to the current ability of the student. In addition, PERM can be trained offline and does not employ non-stationary measures of student ability, making it suitable for transfer between students. We demonstrate PERM's ability to represent the environment parameter space, and training with RL agents with PERM produces a strong performance in deterministic environments. Lastly, we show that our method is transferable between students, without any sacrifice in training quality.
Melody extraction is a core task in music information retrieval, and the estimation of pitch, onset and offset are key sub-tasks in melody extraction. Existing methods have limited accuracy, and work for only one type of data, either single-pitch or multi-pitch. In this paper, we propose a highly accurate method for joint estimation of pitch, onset and offset, named JEPOO. We address the challenges of joint learning optimization and handling both single-pitch and multi-pitch data through novel model design and a new optimization technique named Pareto modulated loss with loss weight regularization. This is the first method that can accurately handle both single-pitch and multi-pitch music data, and even a mix of them. A comprehensive experimental study on a wide range of real datasets shows that JEPOO outperforms state-of-the-art methods by up to 10.6\%, 8.3\% and 10.3\% for the prediction of Pitch, Onset and Offset, respectively, and JEPOO is robust for various types of data and instruments. The ablation study validates the effectiveness of each component of JEPOO.
Factor model is a fundamental investment tool in quantitative investment, which can be empowered by deep learning to become more flexible and efficient in practical complicated investing situations. However, it is still an open question to build a factor model that can conduct stock prediction in an online and adaptive setting, where the model can adapt itself to match the current market regime identified based on only point-in-time market information. To tackle this problem, we propose the first deep learning based online and adaptive factor model, HireVAE, at the core of which is a hierarchical latent space that embeds the underlying relationship between the market situation and stock-wise latent factors, so that HireVAE can effectively estimate useful latent factors given only historical market information and subsequently predict accurate stock returns. Across four commonly used real stock market benchmarks, the proposed HireVAE demonstrate superior performance in terms of active returns over previous methods, verifying the potential of such online and adaptive factor model.
The high rate of false arrhythmia alarms in intensive care units (ICUs) can negatively impact patient care and lead to slow staff response time due to alarm fatigue. To reduce false alarms in ICUs, previous works proposed conventional supervised learning methods which have inherent limitations in dealing with high-dimensional, sparse, unbalanced, and limited data. We propose a deep generative approach based on the conditional denoising diffusion model to detect false arrhythmia alarms in the ICUs. Conditioning on past waveform data of a patient, our approach generates waveform predictions of the patient during an actual arrhythmia event, and uses the distance between the generated and the observed samples to classify the alarm. We design a network with residual links and self-attention mechanism to capture long-term dependencies in signal sequences, and leverage the contrastive learning mechanism to maximize distances between true and false arrhythmia alarms. We demonstrate the effectiveness of our approach on the MIMIC II arrhythmia dataset for detecting false alarms in both retrospective and real-time settings.
Due to the insufficiency of electronic health records (EHR) data utilized in practical diagnosis prediction scenarios, most works are devoted to learning powerful patient representations either from structured EHR data (e.g., temporal medical events, lab test results, etc.) or unstructured data (e.g., clinical notes, etc.). However, synthesizing rich information from both of them still needs to be explored. Firstly, the heterogeneous semantic biases across them heavily hinder the synthesis of representation spaces, which is critical for diagnosis prediction. Secondly, the intermingled quality of partial clinical notes leads to inadequate representations of to-be-predicted patients. Thirdly, typical attention mechanisms mainly focus on aggregating information from similar patients, ignoring important auxiliary information from others. To tackle these challenges, we propose a novel visit sequences-clinical notes joint learning approach, dubbed VecoCare. It performs a Gromov-Wasserstein Distance (GWD)-based contrastive learning task and an adaptive masked language model task in a sequential pre-training manner to reduce heterogeneous semantic biases. After pre-training, VecoCare further aggregates information from both similar and dissimilar patients through a dual-channel retrieval mechanism. We conduct diagnosis prediction experiments on two real-world datasets, which indicates that VecoCare outperforms state-of-the-art approaches. Moreover, the findings discovered by VecoCare are consistent with the medical researches.
News-driven quantitative trading (NQT) has been popularly studied in recent years. Most existing NQT methods are performed in a two-step paradigm, i.e., first analyzing markets by a financial prediction task and then making trading decisions, which is doomed to failure due to the nearly futile financial prediction task. To bypass the financial prediction task, in this paper, we focus on reinforcement learning (RL) based NQT paradigm, which leverages news to make profitable trading decisions directly. In this paper, we propose a novel NQT framework SpotlightTrader based on decision trajectory optimization, which can effectively stitch together a continuous and flexible sequence of trading decisions to maximize profits. In addition, we enhance this framework by constructing a spotlight-driven state trajectory that obeys a stochastic process with irregular abrupt jumps caused by spotlight news. Furthermore, in order to adapt to non-stationary financial markets, we propose an effective training pipeline for this framework, which blends offline pretraining with online finetuning to balance exploration and exploitation effectively during online tradings. Extensive experiments on three real-world datasets demonstrate our proposed model’s superiority over the state-of-the-art NQT methods.
Optimizing molecules with desired properties is a crucial step in de novo drug design. While translation-based methods have achieved initial success, they continue to face the challenge of the “exposure bias” problem. The challenge of preventing the “exposure bias” problem of molecule optimization lies in the need for both positive and negative molecules of contrastive learning. That is because generating positive molecules through data augmentation requires domain-specific knowledge, and randomly sampled negative molecules are easily distinguished from the real molecules. Hence, in this work, we propose a molecule optimization method called GPMO, which leverages a gradient perturbation-based contrastive learning method to prevent the “exposure bias” problem in translation-based molecule optimization. With the assistance of positive and negative molecules, GPMO is able to effectively handle both real and artificial molecules. GPMO is a molecule optimization method that is conditioned on matched molecule pairs for drug discovery. Our empirical studies show that GPMO outperforms the state-of-the- art molecule optimization methods. Furthermore, the negative and positive perturbations improve the robustness of GPMO.
Due to repetitive trial-and-error style interactions between agents and a fixed traffic environment during the policy learning, existing Reinforcement Learning (RL)-based Traffic Signal Control (TSC) methods greatly suffer from long RL training time and poor adaptability of RL agents to other complex traffic environments. To address these problems, we propose a novel Adversarial Inverse Reinforcement Learning (AIRL)-based pre-training method named InitLight, which enables effective initial model generation for TSC agents. Unlike traditional RL-based TSC approaches that train a large number of agents simultaneously for a specific multi-intersection environment, InitLight pre-trains only one single initial model based on multiple single-intersection environments together with their expert trajectories. Since the reward function learned by InitLight can recover ground-truth TSC rewards for different intersections at optimality, the pre-trained agent can be deployed at intersections of any traffic environments as initial models to accelerate subsequent overall global RL training. Comprehensive experimental results show that, the initial model generated by InitLight can not only significantly accelerate the convergence with much fewer episodes, but also own superior generalization ability to accommodate various kinds of complex traffic environments.
The anonymity of online networks makes tackling fraud increasingly costly. Thanks to the superiority of graph representation learning, graph-based fraud detection has made significant progress in recent years. However, upgrading fraudulent strategies produces more advanced and difficult scams. One common strategy is synergistic camouflage —— combining multiple means to deceive others. Existing methods mostly investigate the differences between relations on individual frauds, that neglect the correlation among multi-relation fraudulent behaviors. In this paper, we design several statistics to validate the existence of synergistic camouflage of fraudsters by exploring the correlation among multi-relation interactions. From the perspective of multi-relation, we find two distinctive features of fraudulent behaviors, i.e., alienation and marginalization. Based on the finding, we propose COFRAUD, a correlation-aware fraud detection model, which innovatively incorporates synergistic camouflage into fraud detection. It captures the correlation among multi-relation fraudulent behaviors. Experimental results on two public datasets demonstrate that COFRAUD achieves significant improvements over state-of-the-art methods.