The recent release of large language model (LLM) based chatbots, such as ChatGPT, has attracted significant attention on foundation models. It is widely believed that foundation models will serve as the fundamental building blocks for future AI systems. As foundation models are in their early stages, the design of foundation model based systems has not yet been systematically explored. There is little understanding about the impact of introducing foundation models in software architecture. Therefore, in this paper, we propose a taxonomy of foundation model based systems, which classifies and compares the characteristics of foundation models and design options of foundation model based systems. Our taxonomy comprises three categories: foundation model pretraining and fine-tuning, architecture design of foundation model based systems, and responsible-AI-by-design. This taxonomy provides concrete guidance for making major design decisions when designing foundation model based systems and highlights trade-offs arising from design decisions.

A method for the nonintrusive and structure-preserving model reduction of canonical and noncanonical Hamiltonian systems is presented. Based on the idea of operator inference, this technique is provably convergent and reduces to a straightforward linear solve given snapshot data and gray-box knowledge of the system Hamiltonian. Examples involving several hyperbolic partial differential equations show that the proposed method yields reduced models which, in addition to being accurate and stable with respect to the addition of basis modes, preserve conserved quantities well outside the range of their training data.

Complex Query Answering (CQA) is an important and fundamental task for knowledge graph (KG) reasoning. Query encoding (QE) is proposed as a fast and robust solution to CQA. In the encoding process, most existing QE methods first parse the logical query into an executable computational direct-acyclic graph (DAG), then use neural networks to parameterize the operators, and finally, recursively execute these neuralized operators. However, the parameterization-and-execution paradigm may be potentially over-complicated, as it can be structurally simplified by a single neural network encoder. Meanwhile, sequence encoders, like LSTM and Transformer, proved to be effective for encoding semantic graphs in related tasks. Motivated by this, we propose sequential query encoding (SQE) as an alternative to encode queries for CQA. Instead of parameterizing and executing the computational graph, SQE first uses a search-based algorithm to linearize the computational graph to a sequence of tokens and then uses a sequence encoder to compute its vector representation. Then this vector representation is used as a query embedding to retrieve answers from the embedding space according to similarity scores. Despite its simplicity, SQE demonstrates state-of-the-art neural query encoding performance on FB15k, FB15k-237, and NELL on an extended benchmark including twenty-nine types of in-distribution queries. Further experiment shows that SQE also demonstrates comparable knowledge inference capability on out-of-distribution queries, whose query types are not observed during the training process.

Discourse processing suffers from data sparsity, especially for dialogues. As a result, we explore approaches to build discourse structures for dialogues, based on attention matrices from Pre-trained Language Models (PLMs). We investigate multiple tasks for fine-tuning and show that the dialogue-tailored Sentence Ordering task performs best. To locate and exploit discourse information in PLMs, we propose an unsupervised and a semi-supervised method. Our proposals achieve encouraging results on the STAC corpus, with F1 scores of 57.2 and 59.3 for unsupervised and semi-supervised methods, respectively. When restricted to projective trees, our scores improved to 63.3 and 68.1.

Large language models (LLMs) that have been trained on multilingual but not parallel text exhibit a remarkable ability to translate between languages. We probe this ability in an in-depth study of the pathways language model (PaLM), which has demonstrated the strongest machine translation (MT) performance among similarly-trained LLMs to date. We investigate various strategies for choosing translation examples for few-shot prompting, concluding that example quality is the most important factor. Using optimized prompts, we revisit previous assessments of PaLM's MT capabilities with more recent test sets, modern MT metrics, and human evaluation, and find that its performance, while impressive, still lags that of state-of-the-art supervised systems. We conclude by providing an analysis of PaLM's MT output which reveals some interesting properties and prospects for future work.

Empirical studies suggest a deep intertwining between opinion formation and decision-making processes, but these have been treated as separate problems in the study of dynamical models for social networks. In this paper, we bridge the gap in the literature by proposing a novel coevolutionary model, in which each individual selects an action from a binary set and has an opinion on which action they prefer. Actions and opinions coevolve on a two-layer network. For homogeneous parameters, undirected networks, and under reasonable assumptions on the asynchronous updating mechanics, we prove that the coevolutionary dynamics is an ordinal potential game, enabling analysis via potential game theory. Specifically, we establish global convergence to the Nash equilibria of the game, proving that actions converge in a finite number of time steps, while opinions converge asymptotically. Next, we provide sufficient conditions for the existence of, and convergence to, polarized equilibria, whereby the population splits into two communities, each selecting and supporting one of the actions. Finally, we use simulations to examine the social psychological phenomenon of pluralistic ignorance.

Our group has been developing and operating a platform to acquire, archive, and manage various data related to the Earth's environment to make it available to researchers across a wide range of fields. Development of this system began in the 1980s to receive, archive, and distribute Asian satellite image data. Currently, the system covers a variety of data including weather, climate change, disaster prevention, biodiversity, health, and agriculture. Today, the Data Integration and Analysis System (DIASa) is a large-scale analysis platform with huge storage and more than 10,000 registered users (half of them in Japan and the other half primarily in Asia). As shown in the accompanying figure, users can easily use data collected by the common collection API through the common use API, and they can operate services at the application layer.

Is it possible for a developing country like Vietnam to be a competitive player on the world stage in cutting-edge artificial intelligence (AI) research and development? Will it be able to tap into the $US15.7 trillion projected for the AI global economy by 2030? For Vietnam, these questions often went unchallenged; contemplating answers was daunting. VinAI Research, however, aims to embrace these challenges by laying the groundwork for AI innovation and growth for the region. Founded in 2019, VinAI leapfrogged to the 20th ranking on Thundermark Capital's list of "Global AI Research Companies" by 2022, and was the only Southeast Asian (SEA) representative on the list.a

The New Zealand government identified numerous challenges related to trust and truth in the context of digital technologies. These challenges result from an ever-increasing amount of online social networks, end-to-end digital supply chains, automated decision-making tools, generative artificial intelligence (AI), and cyber-physical systems. Such challenges impact people's lives across professional and private contexts and led to the Veracity Projecta 2021–2024. Outside the field of computing, veracity is not a common term in everyday language. One dictionary definition is "conformity with truth or fact."b

For the world to realize the benefits brought by AI, it is important to ensure artificial intelligent (AI) systems are responsibly developed, used throughout their entire life cycle, and trusted by the humans expected to rely on them.1 The goal for AI adoption has triggered a significant national effort to realize responsible AI (RAI) in Australia. CSIRO Data61 is the data and digital specialist arm of Australia's national science agency. In 2019, CSIRO Data61's worked with the Australian government to conduct the AI Ethics Framework research. This work led to the release of eight AI ethics principles to ensure Australia's adoption of AI is safe, secure, and reliable.a It is challenging to turn high-level AI ethics principles into real-life practices.