Te reo M\=aori (referred to as M\=aori), New Zealand's indigenous language, is under-resourced in language technology. M\=aori speakers are bilingual, where M\=aori is code-switched with English. Unfortunately, there are minimal resources available for M\=aori language technology, language detection and code-switch detection between M\=aori-English pair. Both English and M\=aori use Roman-derived orthography making rule-based systems for detecting language and code-switching restrictive. Most M\=aori language detection is done manually by language experts. This research builds a M\=aori-English bilingual database of 66,016,807 words with word-level language annotation. The New Zealand Parliament Hansard debates reports were used to build the database. The language labels are assigned using language-specific rules and expert manual annotations. Words with the same spelling, but different meanings, exist for M\=aori and English. These words could not be categorised as M\=aori or English based on word-level language rules. Hence, manual annotations were necessary. An analysis reporting the various aspects of the database such as metadata, year-wise analysis, frequently occurring words, sentence length and N-grams is also reported. The database developed here is a valuable tool for future language and speech technology development for Aotearoa New Zealand. The methodology followed to label the database can also be followed by other low-resourced language pairs.

Aspect Sentiment Triplet Extraction (ASTE) aims to extract the spans of aspect, opinion, and their sentiment relations as sentiment triplets. Existing works usually formulate the span detection as a 1D token tagging problem, and model the sentiment recognition with a 2D tagging matrix of token pairs. Moreover, by leveraging the token representation of Pretrained Language Encoders (PLEs) like BERT, they can achieve better performance. However, they simply leverage PLEs as feature extractors to build their modules but never have a deep look at what specific knowledge does PLEs contain. In this paper, we argue that instead of further designing modules to capture the inductive bias of ASTE, PLEs themselves contain "enough" features for 1D and 2D tagging: (1) The token representation contains the contextualized meaning of token itself, so this level feature carries necessary information for 1D tagging. (2) The attention matrix of different PLE layers can further capture multi-level linguistic knowledge existing in token pairs, which benefits 2D tagging. (3) Furthermore, with simple transformations, these two features can also be easily converted to the 2D tagging matrix and 1D tagging sequence, respectively. That will further boost the tagging results. By doing so, PLEs can be natural tagging frameworks and achieve a new state of the art, which is verified by extensive experiments and deep analyses.

The vessel, developed by the British company Sea-Kit International, is surveying the volcano as part of the second phase of the Tonga Eruption Seabed Mapping Project (TESMaP), led by New Zealand's National Institute of Water and Atmospheric Research (Niwa) and funded by the Nippon Foundation of Japan.

Four years ago, the $70 billion Alibaba Group, one of the world's biggest artificial intelligence users, teamed up with Mars, the $35 billion global leader in confectioneries, to figure out the types of candy and chocolates that consumers in China prefer. The fresh consumer data that Alibaba continually gathers from the millions of people shopping on its various platforms turned up the counterintuitive finding that many Chinese who buy chocolates also purchase spicy snacks at the same time. Using that data-driven insight, Mars developed a sweet-and-spicy product: a candy bar that contains Szechuan peppercorns, the source of China's spicy "mala" flavor. Even though Mars didn't conduct any other consumer research to reinforce the A.I.-driven insight, Spicy Snickers proved to be a winner on the mainland. Depending on A.I. also saved the company time; instead of the two to three years that it normally takes to launch a product, Mars was able to bring Spicy Snickers to market for the first time in August 2017, less than 12 months after the collaboration with Alibaba started.

This paper addresses policy learning in non-stationary environments and games with continuous actions. Rather than the classical reward maximization mechanism, inspired by the ideas of follow-the-regularized-leader (FTRL) and mirror descent (MD) update, we propose a no-regret style reinforcement learning algorithm PORL for continuous action tasks. We prove that PORL has a last-iterate convergence guarantee, which is important for adversarial and cooperative games. Empirical studies show that, in stationary environments such as MuJoCo locomotion controlling tasks, PORL performs equally well as, if not better than, the soft actor-critic (SAC) algorithm; in non-stationary environments including dynamical environments, adversarial training, and competitive games, PORL is superior to SAC in both a better final policy performance and a more stable training process.

Robust speech emotion recognition relies on the quality of the speech features. We present speech features enhancement strategy that improves speech emotion recognition. We used the INTERSPEECH 2010 challenge feature-set. We identified subsets from the features set and applied Principle Component Analysis to the subsets. Finally, the features are fused horizontally. The resulting feature set is analyzed using t-distributed neighbour embeddings (t-SNE) before the application of features for emotion recognition. The method is compared with the state-of-the-art methods used in the literature. The empirical evidence is drawn using two well-known datasets: Emotional Speech Dataset (EMO-DB) and Ryerson Audio-Visual Database of Emotional Speech and Song (RAVDESS) for two languages, German and English, respectively. Our method achieved an average recognition gain of 11.5\% for six out of seven emotions for the EMO-DB dataset, and 13.8\% for seven out of eight emotions for the RAVDESS dataset as compared to the baseline study.

We address the problem of unsupervised extractive document summarization, especially for long documents. We model the unsupervised problem as a sparse auto-regression one and approximate the resulting combinatorial problem via a convex, norm-constrained problem. We solve it using a dedicated Frank-Wolfe algorithm. To generate a summary with $k$ sentences, the algorithm only needs to execute $\approx k$ iterations, making it very efficient. We explain how to avoid explicit calculation of the full gradient and how to include sentence embedding information. We evaluate our approach against two other unsupervised methods using both lexical (standard) ROUGE scores, as well as semantic (embedding-based) ones. Our method achieves better results with both datasets and works especially well when combined with embeddings for highly paraphrased summaries.

Stackelberg games are a canonical model for strategic principal-agent interactions. Consider a defense system that distributes its security resources across high-risk targets prior to attacks being executed; or a tax policymaker who sets rules on when audits are triggered prior to seeing filed tax reports; or a seller who chooses a price prior to knowing a customer's proclivity to buy. In each of these scenarios, a principal first selects an action X and then an agent reacts with an action Y, where X and Y are the principal's and agent's action spaces, respectively. In the examples above, agent actions correspond to which target to attack, how much tax to pay to evade an audit, and how much to purchase, respectively. Typically, the principal wants an that maximizes their payoff when the agent plays a best response = br(); such a pair (,) is a Stackelberg equilibrium. By committing to a strategy, the principal can guarantee they achieve a higher payoff than in the fixed point equilibrium of the corresponding simultaneous-play game. However, finding such a strategy requires knowledge of the agent's payoff function. When faced with unknown agent payoffs, the principal can attempt to learn a best response via repeated interactions with the agent. If a (naïve) agent is unaware that such learning occurs and always plays a best response, the principal can use classical online learning approaches to optimize their own payoff in the stage game.

The journal impact factor (JIF) is often equated with journal quality and the quality of the peer review of the papers submitted to the journal. We examined the association between the content of peer review and JIF by analysing 10,000 peer review reports submitted to 1,644 medical and life sciences journals. Two researchers hand-coded a random sample of 2,000 sentences. We then trained machine learning models to classify all 187,240 sentences as contributing or not contributing to content categories. We examined the association between ten groups of journals defined by JIF deciles and the content of peer reviews using linear mixed-effects models, adjusting for the length of the review. The JIF ranged from 0.21 to 74.70. The length of peer reviews increased from the lowest (median number of words 185) to the JIF group (387 words). The proportion of sentences allocated to different content categories varied widely, even within JIF groups. For thoroughness, sentences on 'Materials and Methods' were more common in the highest JIF journals than in the lowest JIF group (difference of 7.8 percentage points; 95% CI 4.9 to 10.7%). The trend for 'Presentation and Reporting' went in the opposite direction, with the highest JIF journals giving less emphasis to such content (difference -8.9%; 95% CI -11.3 to -6.5%). For helpfulness, reviews for higher JIF journals devoted less attention to 'Suggestion and Solution' and provided fewer Examples than lower impact factor journals. No, or only small differences were evident for other content categories. In conclusion, peer review in journals with higher JIF tends to be more thorough in discussing the methods used but less helpful in terms of suggesting solutions and providing examples. Differences were modest and variability high, indicating that the JIF is a bad predictor for the quality of peer review of an individual manuscript.

Natural language understanding (NLU) is the task of semantic decoding of human languages by machines. NLU models rely heavily on large training data to ensure good performance. However, substantial languages and domains have very few data resources and domain experts. It is necessary to overcome the data scarcity challenge, when very few or even zero training samples are available. In this thesis, we focus on developing cross-lingual and cross-domain methods to tackle the low-resource issues. First, we propose to improve the model's cross-lingual ability by focusing on the task-related keywords, enhancing the model's robustness and regularizing the representations. We find that the representations for low-resource languages can be easily and greatly improved by focusing on just the keywords. Second, we present Order-Reduced Modeling methods for the cross-lingual adaptation, and find that modeling partial word orders instead of the whole sequence can improve the robustness of the model against word order differences between languages and task knowledge transfer to low-resource languages. Third, we propose to leverage different levels of domain-related corpora and additional masking of data in the pre-training for the cross-domain adaptation, and discover that more challenging pre-training can better address the domain discrepancy issue in the task knowledge transfer. Finally, we introduce a coarse-to-fine framework, Coach, and a cross-lingual and cross-domain parsing framework, X2Parser. Coach decomposes the representation learning process into a coarse-grained and a fine-grained feature learning, and X2Parser simplifies the hierarchical task structures into flattened ones. We observe that simplifying task structures makes the representation learning more effective for low-resource languages and domains.