AAAI 2023 Papers

Artificial intelligence course has been required to take for compulsory education students in China. However, not all teachers and schools are fully prepared and ready. This is partially because of the lack of adequate teaching and learning resources, which requires a major expenditure of time and effort for schools and teachers to design and develop. To meet the challenge of lacking appropriate resources in teaching and learning AI from grade 1 to grade 9, we developed AI knowledge structure and instructional resources based on Chinese national curriculum for information science and technology. Our comprehensive AI syllabus contains 90 core concepts, 63 learning indicators, and 27 teaching and learning resources, which have been implemented. The resources have been taken as model courses in teacher training programs and an exemplary course has been implemented in primary schools that verified the effectiveness of our resources.

Characters are essential to the plot of any story. Establishing the characters before writing a story can improve the clarity of the plot and the overall flow of the narrative. However, previous work on visual storytelling tends to focus on detecting objects in images and discovering relationships between them. In this approach, characters are not distinguished from other objects when they are fed into the generation pipeline. The result is a coherent sequence of events rather than a character-centric story. In order to address this limitation, we introduce the VIST-Character dataset, which provides rich character-centric annotations, including visual and textual co-reference chains and importance ratings for characters. Based on this dataset, we propose two new tasks: important character detection and character grounding in visual stories. For both tasks, we develop simple, unsupervised models based on distributional similarity and pre-trained vision-and-language models. Our new dataset, together with these models, can serve as the foundation for subsequent work on analysing and generating stories from a character-centric perspective.

The use of voice-over-IP technology has rapidly expanded over the past several years, and has thus become a significant portion of traffic in the real, complex network environment. Deep packet inspection and middlebox technologies need to analyze call flows in order to perform network management, load-balancing, content monitoring, forensic analysis, and intelligence gathering. Because the session setup and management data can be sent on different ports or out of sync with VoIP call data over the Real-time Transport Protocol (RTP) with low latency, inspection software may miss calls or parts of calls. To solve this problem, we engineered two different deep learning models based on hidden representation learning. MAPLE, a matrix-based encoder which transforms packets into an image representation, uses convolutional neural networks to determine RTP packets from data flow. DATE is a density-analysis based tensor encoder which transforms packet data into a three-dimensional point cloud representation. We then perform density-based clustering over the point clouds as latent representations of the data, and classify packets as RTP or non-RTP based on their statistical clustering features. In this research, we show that these tools may allow a data collection and analysis pipeline to begin detecting and buffering RTP streams for later session association, solving the initial drop problem. MAPLE achieves over ninety-nine percent accuracy in RTP/non-RTP detection. The results of our experiments show that both models can not only classify RTP versus non-RTP packet streams, but could extend to other network traffic classification problems in real deployments of network analysis pipelines.

Healthcare acquired infections (HAIs) (e.g., Methicillin-resistant Staphylococcus aureus infection) have complex transmission pathways, spreading not just via direct person-to-person contacts, but also via contaminated surfaces. Prior work in mathematical epidemiology has led to a class of models – which we call load sharing models – that provide a discrete-time, stochastic formalization of HAI-spread on temporal contact networks. The focus of this paper is the source detection problem for the load sharing model. The source detection problem has been studied extensively in SEIR type models, but this prior work does not apply to load sharing models. We show that a natural formulation of the source detection problem for the load sharing model is computationally hard, even to approximate. We then present two alternate formulations that are much more tractable. The tractability of our problems depends crucially on the submodularity of the expected number of infections as a function of the source set. Prior techniques for showing submodularity, such as the "live graph" technique are not applicable for the load sharing model and our key technical contribution is to use a more sophisticated "coupling" technique to show the submodularity result. We propose algorithms for our two problem formulations by extending existing algorithmic results from submodular optimization and combining these with an expectation propagation heuristic for the load sharing model that leads to orders-of-magnitude speedup. We present experimental results on temporal contact networks based on fine-grained EMR data from three different hospitals. Our results on synthetic outbreaks on these networks show that our algorithms outperform baselines by up to 5.97 times. Furthermore, case studies based on hospital outbreaks of Clostridioides difficile infection show that our algorithms identify clinically meaningful sources.

Multivariate time series anomaly detection has been extensively studied under the one-class classification setting, where a training dataset with all normal instances is required. However, preparing such a dataset is very laborious since each single data instance should be fully guaranteed to be normal. It is, therefore, desired to explore multivariate time series anomaly detection methods based on the dataset without any label knowledge. In this paper, we propose MTGFlow, an unsupervised anomaly detection approach forMultivariate Time series anomaly detection via dynamic Graph and entityaware normalizing Flow, leaning only on a widely accepted hypothesis that abnormal instances exhibit sparse densities than the normal. However, the complex interdependencies among entities and the diverse inherent characteristics of each entity pose significant challenges to density estimation, let alone to detect anomalies based on the estimated possibility distribution. To tackle these problems, we propose to learn the mutual and dynamic relations among entities via a graph structure learning model, which helps to model the accurate distribution of multivariate time series. Moreover, taking account of distinct characteristics of the individual entities, an entity-aware normalizing flow is developed to describe each entity into a parameterized normal distribution, thereby producing fine-grained density estimation. Incorporating these two strategies, MTGFlow achieves superior anomaly detection performance. Experiments on five public datasets with seven baselines are conducted, MTGFlow outperforms the SOTA methods by up to 5.0 AUROC%.

Inclusive team participation is one of the most important factors that aids effective collaboration and pair programming. In this paper, we investigated the ability of linguistic features and a transformer-based language model to detect exclusive and inclusive language. The task of detecting exclusive language was approached as a text classification problem. We created a research community resource consisting of a dataset of 40,490 labeled utterances obtained from three programming assignments involving 34 students pair programming in a remote environment. This research involves the first successful automated detection of exclusive language during pair programming. Additionally, this is the first work to perform a computational linguistic analysis on the verbal interaction common in the context of inclusive and exclusive language during pair programming.

The opioid overdose epidemic represents a serious public health crisis, with fatality rates rising considerably over the past several years. To help address the abuse of prescription opioids, state governments collect data on dispensed prescriptions, yet the use of these data is typically limited to manual searches. In this paper, we propose a novel graph-based framework for detecting anomalous opioid prescribing patterns in state Prescription Drug Monitoring Program (PDMP) data, which could aid governments in deterring opioid diversion and abuse. Specifically, we seek to identify connected networks of opioid prescribers and dispensers who engage in high-risk and possibly illicit activity. We develop and apply a novel extension of the Non-Parametric Heterogeneous Graph Scan (NPHGS) to two years of de-identified PDMP data from the state of Kansas, and find that NPHGS identifies subgraphs that are significantly more anomalous than those detected by other graph-based methods. NPHGS also reveals clusters of potentially illicit activity, which may strengthen state law enforcement and regulatory capabilities. Our paper is the first to demonstrate how prescription data can systematically identify anomalous opioid prescribers and dispensers, as well as illustrating the efficacy of a network-based approach. Additionally, our technical extensions to NPHGS offer both improved flexibility and graph density reduction, enabling the framework to be replicated across jurisdictions and extended to other problem domains.

Machine learning and deep learning-based decision making has become part of today's software. The goal of this work is to ensure that machine learning and deep learning-based systems are as trusted as traditional software. Traditional software is made dependable by following rigorous practice like static analysis, testing, debugging, verifying, and repairing throughout the development and maintenance life-cycle. Similarly for machine learning systems, we need to keep these models up to date so that their performance is not compromised. For this, current systems rely on scheduled re-training of these models as new data kicks in. In this work, we propose DetAIL, a tool to measure the data drift that takes place when new data kicks in so that one can adaptively re-train the models whenever re-training is actually required irrespective of schedules. In addition to that, we generate various explanations at sentence level and dataset level to capture why a given payload text has drifted.

Bayesian optimal experimental design is a sub-field of statistics focused on developing methods to make efficient use of experimental resources. Any potential design is evaluated in terms of a utility function, such as the (theoretically well-justified) expected information gain (EIG); unfortunately however, under most circumstances the EIG is intractable to evaluate. In this work we build off of successful variational approaches, which optimize a parameterized variational model with respect to bounds on the EIG. Past work focused on learning a new variational model from scratch for each new design considered. Here we present a novel neural architecture that allows experimenters to optimize a single variational model that can estimate the EIG for potentially infinitely many designs. To further improve computational efficiency, we also propose to train the variational model on a significantly cheaper-to-evaluate lower bound, and show empirically that the resulting model provides an excellent guide for more accurate, but expensive to evaluate bounds on the EIG. We demonstrate the effectiveness of our technique on generalized linear models, a class of statistical models that is widely used in the analysis of controlled experiments. Experiments show that our method is able to greatly improve accuracy over existing approximation strategies, and achieve these results with far better sample efficiency.

Unsupervised depth completion aims to recover dense depth from the sparse one without using the ground-truth annotation. Although depth measurement obtained from LiDAR is usually sparse, it contains valid and real distance information, i.e., scale-consistent absolute depth values. Meanwhile, scale-agnostic counterparts seek to estimate relative depth and have achieved impressive performance. To leverage both the inherent characteristics, we thus suggest to model scale-consistent depth upon unsupervised scale-agnostic frameworks. Specifically, we propose the decomposed scale-consistent learning (DSCL) strategy, which disintegrates the absolute depth into relative depth prediction and global scale estimation, contributing to individual learning benefits. But unfortunately, most existing unsupervised scale-agnostic frameworks heavily suffer from depth holes due to the extremely sparse depth input and weak supervisory signal. To tackle this issue, we introduce the global depth guidance (GDG) module, which attentively propagates dense depth reference into the sparse target via novel dense-to-sparse attention. Extensive experiments show the superiority of our method on outdoor KITTI, ranking 1st and outperforming the best KBNet more than 12% in RMSE. Additionally, our approach achieves state-of-the-art performance on indoor NYUv2 benchmark as well.

Improperly constructed datasets can result in inaccurate inferences. For instance, models trained on biased datasets perform poorly in terms of generalization (i.e., dataset bias). Recent debiasing techniques have successfully achieved generalization performance by underestimating easy-to-learn samples (i.e., bias-aligned samples) and highlighting difficult-to-learn samples (i.e., bias-conflicting samples). However, these techniques may fail owing to noisy labels, because the trained model recognizes noisy labels as difficult-to-learn and thus highlights them. In this study, we find that earlier approaches that used the provided labels to quantify difficulty could be affected by the small proportion of noisy labels. Furthermore, we find that running denoising algorithms before debiasing is ineffective because denoising algorithms reduce the impact of difficult-to-learn samples, including valuable bias-conflicting samples. Therefore, we propose an approach called denoising after entropy-based debiasing, i.e., DENEB, which has three main stages. (1) The prejudice model is trained by emphasizing (bias-aligned, clean) samples, which are selected using a Gaussian Mixture Model. (2) Using the per-sample entropy from the output of the prejudice model, the sampling probability of each sample that is proportional to the entropy is computed. (3) The final model is trained using existing denoising algorithms with the mini-batches constructed by following the computed sampling probability. Compared to existing debiasing and denoising algorithms, our method achieves better debiasing performance on multiple benchmarks.

Quality estimation (QE) aims to assess the quality of machine translations when reference translations are unavailable. QE plays a crucial role in many real-world applications of machine translation. Because labeled QE data are usually limited in scale, recent research, such as DirectQE, pre-trains QE models with pseudo QE data and obtains remarkable performance. However, there tends to be inevitable noise in the pseudo data, hindering models from learning QE accurately. Our study shows that the noise mainly comes from the differences between pseudo and real translation outputs. To handle this problem, we propose CLQE, a denoising pre-training framework for QE based on curriculum learning. More specifically, we propose to measure the degree of noise in the pseudo QE data with some metrics based on statistical or distributional features. With the guidance of these metrics, CLQE gradually pre-trains the QE model using data from cleaner to noisier. Experiments on various benchmarks reveal that CLQE outperforms DirectQE and other strong baselines. We also show that with our framework, pre-training converges faster than directly using the pseudo data. We make our CLQE code available (https://github.com/NJUNLP/njuqe).

Deep Metric Learning (DML) is a group of techniques that aim to measure the similarity between objects through the neural network. Although the number of DML methods has rapidly increased in recent years, most previous studies cannot effectively handle noisy data, which commonly exists in practical applications and often leads to serious performance deterioration. To overcome this limitation, in this paper, we build a connection between noisy samples and hard samples in the framework of self-paced learning, and propose a Balanced Self-Paced Metric Learning (BSPML) algorithm with a denoising multi-similarity formulation, where noisy samples are treated as extremely hard samples and adaptively excluded from the model training by sample weighting. Especially, due to the pairwise relationship and a new balance regularization term, the sub-problem w.r.t. sample weights is a nonconvex quadratic function. To efficiently solve this nonconvex quadratic problem, we propose a doubly stochastic projection coordinate gradient algorithm. Importantly, we theoretically prove the convergence not only for the doubly stochastic projection coordinate gradient algorithm, but also for our BSPML algorithm. Experimental results on several standard data sets demonstrate that our BSPML algorithm has better generalization ability and robustness than the state-of-the-art robust DML approaches.

Evaluation of generative models is mostly based on the comparison between the estimated distribution and the ground truth distribution in a certain feature space. To embed samples into informative features, previous works often use convolutional neural networks optimized for classification, which is criticized by recent studies. Therefore, various feature spaces have been explored to discover alternatives. Among them, a surprising approach is to use a randomly initialized neural network for feature embedding. However, the fundamental basis to employ the random features has not been sufficiently justified. In this paper, we rigorously investigate the feature space of models with random weights in comparison to that of trained models. Furthermore, we provide an empirical evidence to choose networks for random features to obtain consistent and reliable results. Our results indicate that the features from random networks can evaluate generative models well similarly to those from trained networks, and furthermore, the two types of features can be used together in a complementary way.

We provide both empirical and theoretical insights to demystify the gravity well phenomenon in the optimization landscape. We start from describe the problem setup and theoretical results (an escape time lower bound) of the Softmax Gravity Well (SGW) in the literature. Then we move toward the understanding of a recent observation called ASR gravity well. We provide an explanation of why normal distribution with high variance can lead to suboptimal plateaus from an energy function point of view. We also contribute to the empirical insights of curriculum learning by comparison of policy initialization by different normal distributions. Furthermore, we provide the ASR escape time lower bound to understand the ASR gravity well theoretically. Future work includes more specific modeling of the reward as a function of time and quantitative evaluation of normal distribution’s influence on policy initialization.

After the pandemic, artificial intelligence (AI) powered support for mental health care has become increasingly important. The breadth and complexity of significant challenges required to provide adequate care involve: (a) Personalized patient understanding, (b) Safety-constrained and medically validated chatbot patient interactions, and (c) Support for continued feedback-based refinements in design using chatbot-patient interactions. We propose Alleviate, a chatbot designed to assist patients suffering from mental health challenges with personalized care and assist clinicians with understanding their patients better. Alleviate draws from an array of publicly available clinically valid mental-health texts and databases, allowing Alleviate to make medically sound and informed decisions. In addition, Alleviate's modular design and explainable decision-making lends itself to robust and continued feedback-based refinements to its design. In this paper, we explain the different modules of Alleviate and submit a short video demonstrating Alleviate's capabilities to help patients and clinicians understand each other better to facilitate optimal care strategies.

In Federated Learning (FL), models are as fragile as centrally trained models against adversarial examples. However, the adversarial robustness of federated learning remains largely unexplored. This paper casts light on the challenge of adversarial robustness of federated learning. To facilitate a better understanding of the adversarial vulnerability of the existing FL methods, we conduct comprehensive robustness evaluations on various attacks and adversarial training methods. Moreover, we reveal the negative impacts induced by directly adopting adversarial training in FL, which seriously hurts the test accuracy, especially in non-IID settings. In this work, we propose a novel algorithm called Decision Boundary based Federated Adversarial Training (DBFAT), which consists of two components (local re-weighting and global regularization) to improve both accuracy and robustness of FL systems. Extensive experiments on multiple datasets demonstrate that DBFAT consistently outperforms other baselines under both IID and non-IID settings.

Regression-based methods have shown high efficiency and effectiveness for multi-view human mesh recovery. The key components of a typical regressor lie in the feature extraction of input views and the fusion of multi-view features. In this paper, we present Pixel-aligned Feedback Fusion (PaFF) for accurate yet efficient human mesh recovery from multi-view images. PaFF is an iterative regression framework that performs feature extraction and fusion alternately. At each iteration, PaFF extracts pixel-aligned feedback features from each input view according to the reprojection of the current estimation and fuses them together with respect to each vertex of the downsampled mesh. In this way, our regressor can not only perceive the misalignment status of each view from the feedback features but also correct the mesh parameters more effectively based on the feature fusion on mesh vertices. Additionally, our regressor disentangles the global orientation and translation of the body mesh from the estimation of mesh parameters such that the camera parameters of input views can be better utilized in the regression process. The efficacy of our method is validated in the Human3.6M dataset via comprehensive ablation experiments, where PaFF achieves 33.02 MPJPE and brings significant improvements over the previous best solutions by more than 29%. The project page with code and video results can be found at https://kairobo.github.io/PaFF/.

This work presents Z-Mask, an effective and deterministic strategy to improve the adversarial robustness of convolutional networks against physically-realizable adversarial attacks. The presented defense relies on specific Z-score analysis performed on the internal network features to detect and mask the pixels corresponding to adversarial objects in the input image. To this end, spatially contiguous activations are examined in shallow and deep layers to suggest potential adversarial regions. Such proposals are then aggregated through a multi-thresholding mechanism. The effectiveness of Z-Mask is evaluated with an extensive set of experiments carried out on models for semantic segmentation and object detection. The evaluation is performed with both digital patches added to the input images and printed patches in the real world. The results confirm that Z-Mask outperforms the state-of-the-art methods in terms of detection accuracy and overall performance of the networks under attack. Furthermore, Z-Mask preserves its robustness against defense-aware attacks, making it suitable for safe and secure AI applications.