IJCAI 2023 论文列表

Opinion diffusion is a crucial phenomenon in social networks, often underlying the way in which a collection of agents develops a consensus on relevant decisions. Voter models are well-known theoretical models to study opinion spreading in social networks and structured populations. Their simplest version assumes that an updating agent will adopt the opinion of a neighboring agent chosen at random. These models allow us to study, for example, the probability that a certain opinion will fixate into a consensus opinion, as well as the expected time it takes for a consensus opinion to emerge. Standard voter models are oblivious to the opinions held by the agents involved in the opinion adoption process. We propose and study a context-dependent opinion spreading process on an arbitrary social graph, in which the probability that an agent abandons opinion a in favor of opinion b depends on both a and b. We discuss the relations of the model with existing voter models and then derive theoretical results for both the fixation probability and the expected consensus time for two opinions, for both the synchronous and the asynchronous update models.

In this paper, the succinctness of various ML models is studied. To be more precise, the existence of polynomial-time and polynomial-space translations between representation languages for classifiers is investigated. The languages that are considered include decision trees, random forests, several types of boosted trees, binary neural networks, Boolean multilayer perceptrons, and various logical representations of binary classifiers. We provide a complete map indicating for every pair of languages C, C' whether or not a polynomial-time / polynomial-space translation exists from C to C'. We also explain how to take advantage of the resulting map for XAI purposes.

Claims about the interpretability of decision trees can be traced back to the origins of machine learning (ML). Indeed, given some input consistent with a decision tree's path, the explanation for the resulting prediction consists of the features in that path. Moreover, a growing number of works propose the use of decision trees, and of other so-called interpretable models, as a possible solution for deploying ML models in high-risk applications. This paper overviews recent theoretical and practical results which demonstrate that for most decision trees, tree paths exhibit so-called explanation redundancy, in that logically sound explanations can often be significantly more succinct than what the features in the path dictates. More importantly, such decision tree explanations can be computed in polynomial-time, and so can be produced with essentially no effort other than traversing the decision tree. The experimental results, obtained on a large range of publicly available decision trees, support the paper's claims.

This paper focuses on the AI-based damage assessment (ADA) applications that leverage state-of-the-art AI techniques to automatically assess the disaster damage severity using online social media imagery data, which aligns well with the ''disaster risk reduction'' target under United Nations' Sustainable Development Goals (UN SDGs). This paper studies an ADA model generality problem where the objective is to address the limitation of current ADA solutions that are often optimized only for a single disaster event and lack the generality to provide accurate performance across different disaster events. To address this limitation, we work with domain experts and local community stakeholders in disaster response to develop CollabGeneral, a subjective logic-driven crowd-AI collaborative learning framework that integrates AI and crowdsourced human intelligence into a principled learning framework to address the ADA model generality problem. Extensive experiments on four real-world ADA datasets demonstrate that CollabGeneral consistently outperforms the state-of-the-art baselines by significantly improving the ADA model generality across different disasters.

The recent popularity of Wordle has revived interest in guessing games. We develop a general method for finding optimal strategies for guessing games while avoiding an exhaustive search. Our main contribution are several theorems that build towards a general theory to prove optimality of a strategy for a guessing game. This work is developed to apply to any guessing game, but we use Wordle as an example to present concrete results.

We study the problem of fairly allocating a set of indivisible items to a set of agents with additive valuations. Recently, Feige et al. (WINE'21) proved that a maximin share (MMS) allocation exists for all instances with n agents and no more than n + 5 items. Moreover, they proved that an MMS allocation is not guaranteed to exist for instances with 3 agents and at least 9 items, or n ≥ 4 agents and at least 3n + 3 items. In this work, we shrink the gap between these upper and lower bounds for guaranteed existence of MMS allocations. We prove that for any integer c > 0, there exists a number of agents n_c such that an MMS allocation exists for any instance with n ≥ n_c agents and at most n + c items, where n_c ≤ ⌊0.6597^c · c!⌋ for allocation of goods and n_c ≤ ⌊0.7838^c · c!⌋ for chores. Furthermore, we show that for n ≠ 3 agents, all instances with n + 6 goods have an MMS allocation.

Pre-training has marked numerous state of the arts in high-level computer vision, while few attempts have ever been made to investigate how pre-training acts in image processing systems. In this paper, we tailor transformer-based pre-training regimes that boost various low-level tasks. To comprehensively diagnose the influence of pre-training, we design a whole set of principled evaluation tools that uncover its effects on internal representations. The observations demonstrate that pre-training plays strikingly different roles in low-level tasks. For example, pre-training introduces more local information to intermediate layers in super-resolution (SR), yielding significant performance gains, while pre-training hardly affects internal feature representations in denoising, resulting in limited gains. Further, we explore different methods of pre-training, revealing that multi-related-task pre-training is more effective and data-efficient than other alternatives. Finally, we extend our study to varying data scales and model sizes, as well as comparisons between transformers and CNNs. Based on the study, we successfully develop state-of-the-art models for multiple low-level tasks.

We study the problem of generating interesting integer sequences with a combinatorial interpretation. For this we introduce a two-step approach. In the first step, we generate first-order logic sentences which define some combinatorial objects, e.g., undirected graphs, permutations, matchings etc. In the second step, we use algorithms for lifted first-order model counting to generate integer sequences that count the objects encoded by the first-order logic formulas generated in the first step. For instance, if the first-order sentence defines permutations then the generated integer sequence is the sequence of factorial numbers n!. We demonstrate that our approach is able to generate interesting new sequences by showing that a non-negligible fraction of the automatically generated sequences can actually be found in the Online Encyclopaedia of Integer Sequences (OEIS) while generating many other similar sequences which are not present in OEIS and which are potentially interesting. A key technical contribution of our work is the method for generation of first-order logic sentences which is able to drastically prune the space of sentences by discarding large fraction of sentences which would lead to redundant integer sequences.

Prompts have been shown to be an effective method to adapt a frozen Pretrained Language Model (PLM) to perform well on downstream tasks. Prompts can be represented by a human-engineered word sequence or by a learned continuous embedding. In this work, we investigate conditional and compositional differentiable prompting. We propose a new model, Prompt Production System (ProPS), which learns to transform task instructions or input metadata, into continuous prompts that elicit task-specific outputs from the PLM. Our model uses a modular network structure based on our neural formulation of Production Systems, which allows the model to learn discrete rules -- neural functions that learn to specialize in transforming particular prompt input patterns, making it suitable for compositional transfer learning and few-shot learning. We present extensive empirical and theoretical analysis and show that ProPS consistently surpasses other PLM adaptation techniques, and often improves upon fully fine-tuned models, on compositional generalization tasks, controllable summarization and multilingual translation, while needing fewer trainable parameters.

The availability of a large amount of unstructured text has generated interest in utilizing it for future decision-making and developing strategies in various critical domains. Despite some progress, automatically generating accurate reasoning models from the raw text is still an active area of research. Furthermore, most proposed approaches focus on a specific do-main. As such, their suggested transformation methods are usually unreliable when applied to other domains. This research aims to develop a framework, SCANER (Semi-automated CAusal Network Extraction from Raw text), to convert raw text into Causal Bayesian Networks (CBNs). The framework will then be employed in various domains to demonstrate its utilization as a decision-support tool. The preliminary experiments have focused on three domains: political narratives, food insecurity, and medical sciences. The future focus is on developing BNs from political narratives and modifying them through various methods to reduce the level of aggressiveness or extremity in the narratives without causing conflict among the masses or countries.

Total variation distance (TV distance) is a fundamental notion of distance between probability distributions. In this work, we introduce and study the problem of computing the TV distance of two product distributions over the domain {0,1}^n. In particular, we establish the following results. 1. The problem of exactly computing the TV distance of two product distributions is #P-complete. This is in stark contrast with other distance measures such as KL, Chi-square, and Hellinger which tensorize over the marginals leading to efficient algorithms. 2. There is a fully polynomial-time deterministic approximation scheme (FPTAS) for computing the TV distance of two product distributions P and Q where Q is the uniform distribution. This result is extended to the case where Q has a constant number of distinct marginals. In contrast, we show that when P and Q are Bayes net distributions the relative approximation of their TV distance is NP-hard.

Demographic fairness has become a critical objective when developing modern visual models for identity-sensitive applications, such as face attribute recognition (FAR). While great efforts have been made to improve the fairness of the models, the investigation on the adversarial robustness of the fairness (e.g., whether the fairness of the models could still be maintained under potential malicious fairness attacks) is largely ignored. Therefore, this paper explores the adversarial robustness of demographic fairness in FAR applications from both attacking and defending perspectives. In particular, we firstly present a novel fairness attack, who aims at corrupting the demographic fairness of face attribute classifiers. Next, to mitigate the effect of the fairness attack, we design an efficient defense algorithm called robust-fair training. With this defense, face attribute classifiers learn how to combat the bias introduced by the fairness attack. As such, the face attribute classifiers are not only trained to be fair, but the fairness is also robust. Our extensive experimental results show the effectiveness of both our proposed attack and defense methods across various model architectures and FAR applications. We believe our work could be strong baselines for future work on robust-fair AI models.

Sequential decision making is an important field in machine learning, encompassing techniques such as online optimization, structured bandits, and reinforcement learning, which have numerous applications such as recommendation systems, online advertising, conversational agents, and robot learning. However, two key challenges face real-world sequential decision making: the need for adaptable models and the need for safety during both learning and execution. Adaptability refers to the ability of a model to quickly adapt to new and diverse environments, which is especially challenging in environments where feedback is sparse. To address this challenge, we propose using meta reinforcement learning with sub-optimal demonstration data. Safety is also critical in real-world sequential decision making. A model that adheres to safety requirements can avoid dangerous outcomes and ensure the safety of humans and other agents in the environment. We propose an approach based on online convex optimization that ensures safety at every time step. Addressing these challenges can lead to the development of more robust, safe, and adaptable AI systems that can perform a wide range of tasks and operate in a variety of environments.

According to theWorld Health Organization, Pneumoconiosis affects millions of workers globally, with an estimated 260,000 deaths annually. The burden of Pneumoconiosis is particularly high in low-income countries, where occupational safety standards are often inadequate, and the prevalence of the disease is increasing rapidly. The reduced availability of expert medical care in rural areas, where these diseases are more prevalent, further adds to the delayed screening and unfavourable outcomes of the disease. This paper aims to highlight the urgent need for early screening and detection of Pneumoconiosis, given its significant impact on affected individuals, their families, and societies as a whole. With the help of low-cost machine learning models, early screening, detection, and prevention of Pneumoconiosis can help reduce healthcare costs, particularly in low-income countries. In this direction, this research focuses on designing AI solutions for detecting different kinds of Pneumoconiosis from chest X-ray data. This will contribute to the Sustainable Development Goal 3 of ensuring healthy lives and promoting well-being for all at all ages, and present the framework for data collection and algorithm for detecting Pneumoconiosis for early screening. The baseline results show that the existing algorithms are unable to address this challenge. Therefore, it is our assertion that this research will improve state-of-the-art algorithms of segmentation, semantic segmentation, and classification not only for this disease but in general medical image analysis literature.

Although object detection has achieved impressive progress and has a broad impact, it still encounters significant challenges in accurately detecting objects in open-world scenarios. The data in real-world open scenes often exhibit characteristics of limited annotations, such as very few annotated samples or even unknown classes with no annotations. Our studies mainly focus on applying detection to various open scenes and addressing the challenges of sparse samples and unknown classes. The comprehensive research aims to develop more powerful and efficient methods for object detection in the open world, making them more suitable for real-world applications.

Two-stage point-to-box network acts as a critical role in the recent popular 3D Siamese tracking paradigm, which first generates proposals and then predicts corresponding proposal-wise scores. However, such a network suffers from tedious hyper-parameter tuning and task misalignment, limiting the tracking performance. Towards these concerns, we propose a simple yet effective one-stage point-to-box network for point cloud-based 3D single object tracking. It synchronizes 3D proposal generation and center-ness score prediction by a parallel predictor without tedious hyper-parameters. To guide a task-aligned score ranking of proposals, a center-aware focal loss is proposed to supervise the training of the center-ness branch, which enhances the network's discriminative ability to distinguish proposals of different quality. Besides, we design a binary target classifier to identify target-relevant points. By integrating the derived classification scores with the center-ness scores, the resulting network can effectively suppress interference proposals and further mitigate task misalignment. Finally, we present a novel one-stage Siamese tracker OSP2B equipped with the designed network. Extensive experiments on challenging benchmarks including KITTI and Waymo SOT Dataset show that our OSP2B achieves leading performance with a considerable real-time speed.

Large-scale deep learning models contribute to significant performance improvements on varieties of downstream tasks. Current data and model parallelism approaches utilize model replication and partition techniques to support the distributed training of ultra-large models. However, directly deploying these systems often leads to sub-optimal training efficiency due to the complex model architectures and the strict device memory constraints. In this paper, we propose Optimal Sharded Data Parallel (OSDP), an automated parallel training system that combines the advantages from both data and model parallelism. Given the model description and the device information, OSDP makes trade-offs between the memory consumption and the hardware utilization, thus automatically generates the distributed computation graph and maximizes the overall system throughput. In addition, OSDP introduces operator splitting to further alleviate peak memory footprints during training with negligible overheads, which enables the trainability of larger models as well as the higher throughput. Extensive experimental results of OSDP on multiple different kinds of large-scale models demonstrate that the proposed strategy outperforms the state-of-the-art in multiple regards.

The task of extracting overlapping and nested events has received significant attention in recent times, as prior research has primarily focused on extracting flat events, overlooking the intricacies of overlapping and nested occurrences. In this work, we present a new approach to Event Extraction (EE) by reformulating it as an object detection task on a table of token pairs. Our proposed one-stage event extractor, called ODEE, can handle overlapping and nested events. The model is designed with a vertex-based tagging scheme and two auxiliary tasks of predicting the spans and types of event trigger words and argument entities, leveraging the full span information of event elements. Furthermore, in the training stage, we introduce a negative sampling method for table cells to address the imbalance problem of positive and negative table cell tags, meanwhile improving computational efficiency. Empirical evaluations demonstrate that ODEE achieves the state-of-the-art performance on three benchmarks for overlapping and nested EE (i.e., FewFC, Genia11, and Genia13). Furthermore, ODEE outperforms current state-of-the-art methods in terms of both number of parameters and inference speed, indicating its high computational efficiency. To facilitate future research in this area, the codes are publicly available at https://github.com/NingJinzhong/ODEE.

Malnutrition among infants and young children is a pervasive public health concern, particularly in developing countries where resources are limited. Millions of children globally suffer from malnourishment and its complications1. Despite the best efforts of governments and organizations, malnourishment persists and remains a leading cause of morbidity and mortality among children under five. Physical measurements, such as weight, height, middle-upper-arm-circumference (muac), and head circumference are commonly used to assess the nutritional status of children. However, this approach can be resource-intensive and challenging to carry out on a large scale. In this research, we are developing NutriAI, a low-cost solution that leverages small sample size classification approach to detect malnutrition by analyzing 2D images of the subjects in multiple poses. The proposed solution will not only reduce the workload of health workers but also provide a more efficient means of monitoring the nutritional status of children. On the dataset prepared as part of this research, the baseline results highlight that the modern deep learning approaches can facilitate malnutrition detection via anthropometric indicators in the presence of diversity with respect to age, gender, physical characteristics, and accessories including clothing.