Cognitive Diagnosis Models (CDMs) are a special family of discrete latent variable models that are widely used in modern educational, psychological, social and biological sciences. A key component of CDMs is a binary $Q$-matrix characterizing the dependence structure between the items and the latent attributes. Additionally, researchers also assume in many applications certain hierarchical structures among the latent attributes to characterize their dependence. In most CDM applications, the attribute-attribute hierarchical structures, the item-attribute $Q$-matrix, the item-level diagnostic model, as well as the number of latent attributes, need to be fully or partially pre-specified, which however may be subjective and misspecified as noted by many recent studies. This paper considers the problem of jointly learning these latent and hierarchical structures in CDMs from observed data with minimal model assumptions. Specifically, a penalized likelihood approach is proposed to select the number of attributes and estimate the latent and hierarchical structures simultaneously. An efficient expectation-maximization (EM) algorithm and a latent structure recovery algorithm are developed, and statistical consistency theory is also established under mild conditions. The good performance of the proposed method is illustrated by simulation studies and a real data application in educational assessment.

The predictive advantage of combining several different predictive models is widely accepted. Particularly in time series forecasting problems, this combination is often dynamic to cope with potential non-stationary sources of variation present in the data. Despite their superior predictive performance, ensemble methods entail two main limitations: high computational costs and lack of transparency. These issues often preclude the deployment of such approaches, in favour of simpler yet more efficient and reliable ones. In this paper, we leverage the idea of model compression to address this problem in time series forecasting tasks. Model compression approaches have been mostly unexplored for forecasting. Their application in time series is challenging due to the evolving nature of the data. Further, while the literature focuses on neural networks, we apply model compression to distinct types of methods. In an extensive set of experiments, we show that compressing dynamic forecasting ensembles into an individual model leads to a comparable predictive performance and a drastic reduction in computational costs. Further, the compressed individual model with best average rank is a rule-based regression model. Thus, model compression also leads to benefits in terms of model interpretability. The experiments carried in this paper are fully reproducible.

The availability of abundant labeled data in recent years led the researchers to introduce a methodology called transfer learning, which utilizes existing data in situations where there are difficulties in collecting new annotated data. Transfer learning aims to boost the performance of a target learner by applying another related source data. In contrast to the traditional machine learning and data mining techniques, which assume that the training and testing data lie from the same feature space and distribution, transfer learning can handle situations where there is a discrepancy between domains and distributions. These characteristics give the model the potential to utilize the available related source data and extend the underlying knowledge to the target task achieving better performance. This survey paper aims to give a concise review of traditional and current transfer learning settings, existing challenges, and related approaches.

Commonsense knowledge acquisition and reasoning have long been a core artificial intelligence problem. However, in the past, there has been a lack of scalable methods to collect commonsense knowledge. In this paper, we propose to develop principles for collecting commonsense knowledge based on selectional preference. We generalize the definition of selectional preference from one-hop linguistic syntactic relations to higher-order relations over linguistic graphs. Unlike previous commonsense knowledge definition (e.g., ConceptNet), the selectional preference (SP) knowledge only relies on statistical distribution over linguistic graphs, which can be efficiently and accurately acquired from the unlabeled corpus with modern tools. Following this principle, we develop a large-scale eventuality (a linguistic term covering activity, state, and event)-based knowledge graph ASER, where each eventuality is represented as a dependency graph, and the relation between them is a discourse relation defined in shallow discourse parsing. The higher-order selectional preference over collected linguistic graphs reflects various kinds of commonsense knowledge. Moreover, motivated by the observation that humans understand events by abstracting the observed events to a higher level and can thus transferring their knowledge to new events, we propose a conceptualization module to significantly boost the coverage of ASER. In total, ASER contains 438 million eventualities and 648 million edges between eventualities. After conceptualization with Probase, a selectional preference based concept-instance relational knowledge base, our concept graph contains 15 million conceptualized eventualities and 224 million edges between them. Detailed analysis is provided to demonstrate its quality. All the collected data, APIs, and tools are available at https://github.com/HKUST-KnowComp/ASER.

Reasoning about information from multiple parts of a passage to derive an answer is an open challenge for reading-comprehension models. In this paper, we present an approach that reasons about complex questions by decomposing them to simpler subquestions that can take advantage of single-span extraction reading-comprehension models, and derives the final answer according to instructions in a predefined reasoning template. We focus on subtraction-based arithmetic questions and evaluate our approach on a subset of the DROP dataset. We show that our approach is competitive with the state-of-the-art while being interpretable and requires little supervision

Despite exciting progress in pre-training for visual-linguistic (VL) representations, very few aspire to a small VL model. In this paper, we study knowledge distillation (KD) to effectively compress a transformer-based large VL model into a small VL model. The major challenge arises from the inconsistent regional visual tokens extracted from different detectors of Teacher and Student, resulting in the misalignment of hidden representations and attention distributions. To address the problem, we retrain and adapt the Teacher by using the same region proposals from Student's detector while the features are from Teacher's own object detector. With aligned network inputs, the adapted Teacher is capable of transferring the knowledge through the intermediate representations. Specifically, we use the mean square error loss to mimic the attention distribution inside the transformer block and present a token-wise noise contrastive loss to align the hidden state by contrasting with negative representations stored in a sample queue. To this end, we show that our proposed distillation significantly improves the performance of small VL models on image captioning and visual question answering tasks. It reaches 120.8 in CIDEr score on COCO captioning, an improvement of 5.1 over its non-distilled counterpart; and an accuracy of 69.8 on VQA 2.0, a 0.8 gain from the baseline. Our extensive experiments and ablations confirm the effectiveness of VL distillation in both pre-training and fine-tuning stages.

Student mobility or academic mobility involves students moving between institutions during their post-secondary education, and one of the challenging tasks in this process is to assess the transfer credits to be offered to the incoming student. In general, this process involves domain experts comparing the learning outcomes of the courses, to decide on offering transfer credits to the incoming students. This manual implementation is not only labor-intensive but also influenced by undue bias and administrative complexity. The proposed research article focuses on identifying a model that exploits the advancements in the field of Natural Language Processing (NLP) to effectively automate this process. Given the unique structure, domain specificity, and complexity of learning outcomes (LOs), a need for designing a tailor-made model arises. The proposed model uses a clustering-inspired methodology based on knowledge-based semantic similarity measures to assess the taxonomic similarity of LOs and a transformer-based semantic similarity model to assess the semantic similarity of the LOs. The similarity between LOs is further aggregated to form course to course similarity. Due to the lack of quality benchmark datasets, a new benchmark dataset containing seven course-to-course similarity measures is proposed. Understanding the inherent need for flexibility in the decision-making process the aggregation part of the model offers tunable parameters to accommodate different scenarios. While providing an efficient model to assess the similarity between courses with existing resources, this research work steers future research attempts to apply NLP in the field of articulation in an ideal direction by highlighting the persisting research gaps.

The dream of building machines that can do science has inspired scientists for decades. Remarkable advances have been made recently; however, we are still far from achieving this goal. In this paper, we focus on the scientific discovery process where a high level of reasoning and remarkable problem-solving ability are required. We review different machine learning techniques used in scientific discovery with their limitations. We survey and discuss the main principles driving the scientific discovery process. These principles are used in different fields and by different scientists to solve problems and discover new knowledge. We provide many examples of the use of these principles in different fields such as physics, mathematics, and biology. We also review AI systems that attempt to implement some of these principles. We argue that building science discovery machines should be guided by these principles as an alternative to the dominant approach of current AI systems that focuses on narrow objectives. Building machines that fully incorporate these principles in an automated way might open the doors for many advancements.

Artificial Intelligence (AI) covers a broad spectrum of computational problems and use cases. Many of those implicate profound and sometimes intricate questions of how humans interact or should interact with AIs. Moreover, many users or future users do have abstract ideas of what AI is, significantly depending on the specific embodiment of AI applications. Human-centered-design approaches would suggest evaluating the impact of different embodiments on human perception of and interaction with AI. An approach that is difficult to realize due to the sheer complexity of application fields and embodiments in reality. However, here XR opens new possibilities to research human-AI interactions. The article's contribution is twofold: First, it provides a theoretical treatment and model of human-AI interaction based on an XR-AI continuum as a framework for and a perspective of different approaches of XR-AI combinations. It motivates XR-AI combinations as a method to learn about the effects of prospective human-AI interfaces and shows why the combination of XR and AI fruitfully contributes to a valid and systematic investigation of human-AI interactions and interfaces. Second, the article provides two exemplary experiments investigating the aforementioned approach for two distinct AI-systems. The first experiment reveals an interesting gender effect in human-robot interaction, while the second experiment reveals an Eliza effect of a recommender system. Here the article introduces two paradigmatic implementations of the proposed XR testbed for human-AI interactions and interfaces and shows how a valid and systematic investigation can be conducted. In sum, the article opens new perspectives on how XR benefits human-centered AI design and development.

With so many interesting machine learning models to learn about and problems to solve it can be easy to forget about one of the core skills you actually need to be an effective data scientist. The vast majority of data scientists are going to have end-to-end responsibility for owning their own data pipelines and sourcing their own data. Also -- you should never have to rely on someone else to pull data for you. It might now be one of the "sexy" skills that are associated with the job, but it is essential (trust me on this one, in FAANG companies you will likely be tested on SQL at interview as an initial screening task). This is perhaps a little more controversial.