Large language models (LLMs), such as ChatGPT/GPT-4, have proven to be powerful tools in promoting the user experience as an AI assistant. The continuous works are proposing multi-modal large language models (MLLM), empowering LLMs with the ability to sense multiple modality inputs through constructing a joint semantic space (e.g. visual-text space). Though significant success was achieved in LLMs and MLLMs, exploring LLMs and MLLMs in domain-specific applications that required domain-specific knowledge and expertise has been less conducted, especially for \textbf{marine domain}. Different from general-purpose MLLMs, the marine-specific MLLM is required to yield much more \textbf{sensitive}, \textbf{informative}, and \textbf{scientific} responses. In this work, we demonstrate that the existing MLLMs optimized on huge amounts of readily available general-purpose training data show a minimal ability to understand domain-specific intents and then generate informative and satisfactory responses. To address these issues, we propose \textbf{MarineGPT}, the first vision-language model specially designed for the marine domain, unlocking the secrets of the ocean to the public. We present our \textbf{Marine-5M} dataset with more than 5 million marine image-text pairs to inject domain-specific marine knowledge into our model and achieve better marine vision and language alignment. Our MarineGPT not only pushes the boundaries of marine understanding to the general public but also offers a standard protocol for adapting a general-purpose assistant to downstream domain-specific experts. We pave the way for a wide range of marine applications while setting valuable data and pre-trained models for future research in both academic and industrial communities.

Despite the impressive growth of the abilities of multilingual language models, such as XLM-R and mT5, it has been shown that they still face difficulties when tackling typologically-distant languages, particularly in the low-resource setting. One obstacle for effective cross-lingual transfer is variability in word-order patterns. It can be potentially mitigated via source- or target-side word reordering, and numerous approaches to reordering have been proposed. However, they rely on language-specific rules, work on the level of POS tags, or only target the main clause, leaving subordinate clauses intact. To address these limitations, we present a new powerful reordering method, defined in terms of Universal Dependencies, that is able to learn fine-grained word-order patterns conditioned on the syntactic context from a small amount of annotated data and can be applied at all levels of the syntactic tree. We conduct experiments on a diverse set of tasks and show that our method consistently outperforms strong baselines over different language pairs and model architectures. This performance advantage holds true in both zero-shot and few-shot scenarios.

While there has been extensive work on deep neural networks for images and text, deep learning for relational databases (RDBs) is still a rather unexplored field. One direction that recently gained traction is to apply Graph Neural Networks (GNNs) to RBDs. However, training GNNs on large relational databases (i.e., data stored in multiple database tables) is rather inefficient due to multiple rounds of training and potentially large and inefficient representations. Hence, in this paper we propose SPARE (Single-Pass Relational models), a new class of neural models that can be trained efficiently on RDBs while providing similar accuracies as GNNs. For enabling efficient training, different from GNNs, SPARE makes use of the fact that data in RDBs has a regular structure, which allows one to train these models in a single pass while exploiting symmetries at the same time. Our extensive empirical evaluation demonstrates that SPARE can significantly speedup both training and inference while offering competitive predictive performance over numerous baselines.

The self-improving ability of large language models (LLMs), enabled by prompting them to analyze and revise their own outputs, has garnered significant interest in recent research. However, this ability has been shown to be absent and difficult to learn for smaller models, thus widening the performance gap between state-of-the-art LLMs and more cost-effective and faster ones. To reduce this gap, we introduce TriPosT, a training algorithm that endows smaller models with such self-improvement ability, and show that our approach can improve a LLaMA-7b's performance on math and reasoning tasks by up to 7.13%. In contrast to prior work, we achieve this by using the smaller model to interact with LLMs to collect feedback and improvements on its own generations. We then replay this experience to train the small model. Our experiments on four math and reasoning datasets show that the interactive experience of learning from and correcting its own mistakes is crucial for small models to improve their performance.

A social robot acting as a 'mediator' can enhance interactions between humans, for example, in fields such as education and healthcare. A particularly promising area of research is the use of a social robot mediator in a multiparty setting, which tends to be the most applicable in real-world scenarios. However, research in social robot mediation for multiparty interactions is still emerging and faces numerous challenges. This paper provides an overview of social robotics and mediation research by highlighting relevant literature and some of the ongoing problems. The importance of incorporating relevant psychological principles for developing social robot mediators is also presented. Additionally, the potential of implementing a Theory of Mind in a social robot mediator is explored, given how such a framework could greatly improve mediation by reading the individual and group mental states to interact effectively.

Quality classification of wood boards is an essential task in the sawmill industry, which is still usually performed by human operators in small to median companies in developing countries. Machine learning algorithms have been successfully employed to investigate the problem, offering a more affordable alternative compared to other solutions. However, such approaches usually present some drawbacks regarding the proper selection of their hyperparameters. Moreover, the models are susceptible to the features extracted from wood board images, which influence the induction of the model and, consequently, its generalization power. Therefore, in this paper, we investigate the problem of simultaneously tuning the hyperparameters of an artificial neural network (ANN) as well as selecting a subset of characteristics that better describes the wood board quality. Experiments were conducted over a private dataset composed of images obtained from a sawmill industry and described using different feature descriptors. The predictive performance of the model was compared against five baseline methods as well as a random search, performing either ANN hyperparameter tuning and feature selection. Experimental results suggest that hyperparameters should be adjusted according to the feature set, or the features should be selected considering the hyperparameter values. In summary, the best predictive performance, i.e., a balanced accuracy of $0.80$, was achieved in two distinct scenarios: (i) performing only feature selection, and (ii) performing both tasks concomitantly. Thus, we suggest that at least one of the two approaches should be considered in the context of industrial applications.

Livestock health and welfare monitoring has traditionally been a labor-intensive task performed manually. Recent advances have led to the adoption of AI and computer vision techniques, particularly deep learning models, as decision-making tools within the livestock industry. These models have been employed for tasks like animal identification, tracking, body part recognition, and species classification. In the past decade, there has been a growing interest in using these models to explore the connection between livestock behaviour and health issues. While previous review studies have been rather generic, there is currently no review study specifically focusing on DL for livestock behaviour recognition. Hence, this systematic literature review (SLR) was conducted. The SLR involved an initial search across electronic databases, resulting in 1101 publications. After applying defined selection criteria, 126 publications were shortlisted. These publications were further filtered based on quality criteria, resulting in the selection of 44 high-quality primary studies. These studies were analysed to address the research questions. The results showed that DL successfully addressed 13 behaviour recognition problems encompassing 44 different behaviour classes. A variety of DL models and networks were employed, with CNN, Faster R-CNN, YOLOv5, and YOLOv4 being among the most common models, and VGG16, CSPDarknet53, GoogLeNet, ResNet101, and ResNet50 being popular networks. Performance evaluation involved ten different matrices, with precision and accuracy being the most frequently used. Primary studies identified challenges, including occlusion, adhesion, data imbalance, and the complexities of the livestock environment. The SLR study also discussed potential solutions and research directions to facilitate the development of autonomous livestock behaviour recognition systems.

One of the most popular clustering algorithms is the celebrated $D^\alpha$ seeding algorithm (also know as $k$-means++ when $\alpha=2$) by Arthur and Vassilvitskii (2007), who showed that it guarantees in expectation an $O(2^{2\alpha}\cdot \log k)$-approximate solution to the ($k$,$\alpha$)-means cost (where euclidean distances are raised to the power $\alpha$) for any $\alpha\ge 1$. More recently, Balcan, Dick, and White (2018) observed experimentally that using $D^\alpha$ seeding with $\alpha>2$ can lead to a better solution with respect to the standard $k$-means objective (i.e. the $(k,2)$-means cost). In this paper, we provide a rigorous understanding of this phenomenon. For any $\alpha>2$, we show that $D^\alpha$ seeding guarantees in expectation an approximation factor of $$ O_\alpha \left((g_\alpha)^{2/\alpha}\cdot \left(\frac{\sigma_{\mathrm{max}}}{\sigma_{\mathrm{min}}}\right)^{2-4/\alpha}\cdot (\min\{\ell,\log k\})^{2/\alpha}\right)$$ with respect to the standard $k$-means cost of any underlying clustering; where $g_\alpha$ is a parameter capturing the concentration of the points in each cluster, $\sigma_{\mathrm{max}}$ and $\sigma_{\mathrm{min}}$ are the maximum and minimum standard deviation of the clusters around their means, and $\ell$ is the number of distinct mixing weights in the underlying clustering (after rounding them to the nearest power of $2$). We complement these results by some lower bounds showing that the dependency on $g_\alpha$ and $\sigma_{\mathrm{max}}/\sigma_{\mathrm{min}}$ is tight. Finally, we provide an experimental confirmation of the effects of the aforementioned parameters when using $D^\alpha$ seeding. Further, we corroborate the observation that $\alpha>2$ can indeed improve the $k$-means cost compared to $D^2$ seeding, and that this advantage remains even if we run Lloyd's algorithm after the seeding.

In the field of natural language processing, open-domain chatbots have emerged as an important research topic. However, a major limitation of existing open-domain chatbot research is its singular focus on short single-session dialogue, neglecting the potential need for understanding contextual information in multiple consecutive sessions that precede an ongoing dialogue. Among the elements that compose the context in multi-session conversation settings, the time intervals between sessions and the relationships between speakers would be particularly important. Despite their importance, current research efforts have not sufficiently addressed these dialogical components. In this paper, we introduce a new 1M multi-session dialogue dataset, called Conversation Chronicles, for implementing a long-term conversation setup in which time intervals and fine-grained speaker relationships are incorporated. Following recent works, we exploit a large language model to produce the data. The extensive human evaluation shows that dialogue episodes in Conversation Chronicles reflect those properties while maintaining coherent and consistent interactions across all the sessions. We also propose a dialogue model, called ReBot, which consists of chronological summarization and dialogue generation modules using only around 630M parameters. When trained on Conversation Chronicles, ReBot demonstrates long-term context understanding with a high human engagement score.

For decision-making, one common approach first The Conversational Machine Reading (CMR) task segments the document into many text spans at (Saeidi et al., 2018) requires an agent to answer an different granularity levels (e.g., sentences or Elementary initial question from users through multi-turn dialogue Discourse Units (EDUs)). Then complex interactions based on a given document. As modules are adopted to predict the entailment state shown in Figure 1, a typical process involves two for each document span based on user scenario and steps, (1) the agent first makes a decision classification previous dialogue history (both are user-provided among IRRELEVANT, YES, NO and MORE, information). Finally, decisions are made based on (2) if the decision is MORE, the agent generates a the entailment states of all document spans. One question to clarify an unmentioned condition in the effective module for predicting entailment states is given document, otherwise responds directly. Recent transformer blocks (Vaswani et al., 2017), which research (Verma et al., 2020; Lawrence et al., are widely adopted (Gao et al., 2020b; Ouyang 2019; Zhong and Zettlemoyer, 2019; Gao et al., et al., 2021; Zhang et al., 2022). However, the 2020a; Gao et al., 2020b; Ouyang et al., 2021; aforementioned approach has overlooked the explicit Zhang et al., 2022) has explored how to improve alignment between the document and the userprovided the abilities of decision-making and question generation.