To enhance English language learning through descriptive tasks based on everyday life scenes, we propose leveraging multimodal Al technologies. This involves utilizing multimodal large language models to automatically assess the quality of student responses and provide personalized, timely feedback in the form of Al-driven comments and suggestions for improvement. Furthermore, we will investigate students' perceptions of the feedback provided and its effectiveness in enhancing their writing skills. To achieve this, 30 participants were recruited to describe a set of daily-life pictures with the Al-driven automated assistance. During the experiment, eye-tracking technology was employed to capture students' eye movement data, enabling analysis of their visual perception and attention allocation. A survey questionnaire was administered to gather students' perceptions of this language learning approach and the effectiveness of the feedback. The experimental results indicate that students exhibit a positive attitude towards this language learning approach, as evidenced by their high levels of learning interest and motivation. Moreover, students demonstrate a willingness to actively engage with automatic feedback, with a particular inclination towards investing more attention and time in the suggestions generated by a large language model.

Large-scale benchmark datasets are crucial in advancing research within the computer science communities. They enable the development of more sophisticated AI models and serve as "golden" benchmarks for evaluating their performance. Thus, ensuring the quality of these datasets is of utmost importance for academic research and the progress of AI systems. For the emerging vision-language tasks, some datasets have been created and frequently used, such as Flickr30k, COCO, and NoCaps, which typically contain a large number of images paired with their ground-truth textual descriptions. In this paper, an automatic method is proposed to assess the quality of large-scale benchmark datasets designed for vision-language tasks. In particular, a new cross-modal matching model is developed, which is capable of automatically scoring the textual descriptions of visual images. Subsequently, this model is employed to evaluate the quality of vision-language datasets by automatically assigning a score to each 'ground-truth' description for every image picture. With a good agreement between manual and automated scoring results on the datasets, our findings reveal significant disparities in the quality of the ground-truth descriptions included in the benchmark datasets. Even more surprising, it is evident that a small portion of the descriptions are unsuitable for serving as reliable ground-truth references. These discoveries emphasize the need for careful utilization of these publicly accessible benchmark databases.

Transformer-based pre-trained language models like BERT and its variants have recently achieved promising performance in various natural language processing (NLP) tasks. However, the conventional paradigm constructs the backbone by purely stacking the manually designed global self-attention layers, introducing inductive bias and thus leads to sub-optimal. In this work, we make the first attempt to automatically discover novel pre-trained language model (PLM) backbone on a flexible search space containing the most fundamental operations from scratch. Specifically, we propose a well-designed search space which (i) contains primitive math operations in the intra-layer level to explore novel attention structures, and (ii) leverages convolution blocks to be the supplementary for attentions in the inter-layer level to better learn local dependency. To enhance the efficiency for finding promising architectures, we propose an Operation-Priority Neural Architecture Search (OP-NAS) algorithm, which optimizes both the search algorithm and evaluation of candidate models. Specifically, we propose Operation-Priority (OP) evolution strategy to facilitate model search via balancing exploration and exploitation. Furthermore, we design a Bi-branch Weight-Sharing (BIWS) training strategy for fast model evaluation. Extensive experiments show that the searched architecture (named AutoBERT-Zero) significantly outperforms BERT and its variants of different model capacities in various downstream tasks, proving the architecture's transfer and scaling abilities. Remarkably, AutoBERT-Zero-base outperforms RoBERTa-base (using much more data) and BERT-large (with much larger model size) by 2.4 and 1.4 higher score on GLUE test set.

Image captioning has emerged as an interesting research field in recent years due to its broad application scenarios. The traditional paradigm of image captioning relies on paired image-caption datasets to train the model in a supervised manner. However, creating such paired datasets for every target language is prohibitively expensive, which hinders the extensibility of captioning technology and deprives a large part of the world population of its benefit. In this work, we present a novel unpaired cross-lingual method to generate image captions without relying on any caption corpus in the source or the target language. Specifically, our method consists of two phases: (1) a cross-lingual auto-encoding process, which utilizing a sentence parallel (bitext) corpus to learn the mapping from the source to the target language in the scene graph encoding space and decode sentences in the target language, and (2) a cross-modal unsupervised feature mapping, which seeks to map the encoded scene graph features from image modality to language modality. We verify the effectiveness of our proposed method on the Chinese image caption generation task. The comparisons against several existing methods demonstrate the effectiveness of our approach.

Technical trading rules have been utilized in the stock markets to make profit for more than a century. However, no single trading rule can ever be expected to predict the stock price trend accurately. In fact, many investors and fund managers make trading decisions by combining a bunch of technical indicators. In this paper, we consider the complex stock trading strategy, called Performance-based Reward Strategy (PRS), proposed by [1]. Instead of combining two classes of technical trading rules, we expand the scope to combine the seven most popular classes of trading rules in financial markets, resulting in a total of 1059 component trading rules. Each component rule is assigned a starting weight and a reward/penalty mechanism based on rules' recent profit is proposed to update their weights over time. To determine the best parameter values of PRS, we employ an improved time variant particle swarm optimization (TVPSO) algorithm with the objective of maximizing the annual net profit generated by PRS. Due to a large number of component rules and swarm size, the optimization time is significant. A parallel PSO based on Hadoop, an open source parallel programming model of MapReduce, is employed to optimize PRS more efficiently. The experimental results show that PRS outperforms all of the component rules in the testing period.

Trading rules have been utilized in the stock market to make profit for more than a century. However, only using a single trading rule may not be sufficient to predict the stock price trend accurately. Although some complex trading strategies combining various classes of trading rules have been proposed in the literature, they often pick only one rule for each class, which may lose valuable information from other rules in the same class. In this paper, a complex stock trading strategy, namely weight reward strategy (WRS), is proposed. WRS combines the two most popular classes of trading rules-moving average (MA) and trading range break-out (TRB). For both MA and TRB, WRS includes different combinations of the rule parameters to get a universe of 140 component trading rules in all. Each component rule is assigned a start weight and a reward/penalty mechanism based on profit is proposed to update these rules' weights over time. To determine the best parameter values of WRS, we employ an improved time variant Particle Swarm Optimization (PSO) algorithm with the objective of maximizing the annual net profit generated by WRS. The experiments show that our proposed WRS optimized by PSO outperforms the best moving average and trading range break-out rules.

Probabilistic principal component analysis (PPCA) is a popular linear latent variable model for performing dimension reduction on 1-D data in a probabilistic manner. However, when used on 2-D data such as images, PPCA suffers from the curse of dimensionality due to the subsequently large number of model parameters. To overcome this problem, we propose in this paper a novel probabilistic model on 2-D data called bilinear PPCA (BPPCA). This allows the establishment of a closer tie between BPPCA and its nonprobabilistic counterpart. Moreover, two efficient parameter estimation algorithms for fitting BPPCA are also developed. Experiments on a number of 2-D synthetic and real-world data sets show that BPPCA is more accurate than existing probabilistic and nonprobabilistic dimension reduction methods.

Accumulator is a highly path dependant derivative structure that has been introduced as a retail financial product in recent years and becomes very popular in some Asian cities with its speculative nature. Despite its popularity, its pricing formula is not well known especially when there is a barrier structure. When the barrier in an accumulator contract is applied continuously, this paper obtains exact analytic pricing formulae for immediate settlement and for delay settlement. For discrete barrier, we also obtain analytic formulae which can approximate the fair price of an accumulator under both settlement methods. Through Monte Carlo simulation, we show that the approximation is highly satisfactory. With price formulae in close forms, this paper further explains how to price the product fairly to fit into its zero-cost structure. The analytic formulae also help in computing the Greeks of an accumulator which are documented in this paper. An asymmetry can be observed here that when the buyer is suffering a loss, risk characteristics like delta and vega are substantially larger than when the buyer is enjoying a profit. This means that losing buyers will be more vulnerable to price changes and volatility changes than winning buyers. This is consistent with another observation in the paper that the value at risk for the buyer can be several times larger than that of the seller.

In this brief, we propose a fast expectation conditional maximization (ECM) algorithm for maximum-likelihood (ML) estimation of mixtures of factor analyzers (MFA). Unlike the existing expectation-maximization (EM) algorithms such as the EM in Ghahramani and Hinton, 1996, and the alternating ECM (AECM) in McLachlan and Peel, 2003, where the missing data contains component-indicator vectors as well as latent factors, the missing data in our ECM consists of component-indicator vectors only. The novelty of our algorithm is that closed-form expressions in all conditional maximization (CM) steps are obtained explicitly, instead of resorting to numerical optimization methods. As revealed by experiments, the convergence of our ECM is substantially faster than EM and AECM regardless of whether assessed by central processing unit (CPU) time or number of iterations.

We suggest using independent component analysis (ICA) to decompose multivariate time series into statistically independent time series. Then, we propose to use ICA-GARCH models which are computationally efficient to estimate the multivariate volatilities. The experimental results show that the ICA-GARCH models are more effective than existing methods, including DCC, PCA-GARCH, and EWMA. We also apply the proposed models to compute value at risk (VaR) for risk management applications. The backtesting and the out-of-sample tests validate the performance of ICA-GARCH models for value at risk estimation.

Term structure is a useful curve describing some financial asset as a function of time to maturity or expiration. In this paper, we propose to use Independent Component Analysis (ICA) to model the term structure of multiple yield curves. The idea is that we first employ ICA to decompose the multivariate time series, then we suggest two ICA methods for dimension reduction and pattern recognition of the term structure. We also compare the results by using an alternative method, Principal Component Analysis (PCA). The empirical studies suggest that the proposed ICA approaches outperform PCA methods in modeling the term structure. This model can be used in financial time series analysis as well as related financial applications.

OBJECTIVE: Given the slow adoption of medical informatics in Hong Kong and Asia, we sought to understand the contributory barriers and potential incentives associated with information technology implementation. DESIGN AND MEASUREMENTS: A representative sample of 949 doctors (response rate = 77.0%) was asked through a postal survey to rank a list of nine barriers associated with clinical computerization according to self-perceived importance. They ranked seven incentives or catalysts that may influence computerization. We generated mean rank scores and used multidimensional preference analysis to explore key explanatory dimensions of these variables. A hierarchical cluster analysis was performed to identify homogenous subgroups of respondents. We further determined the relationships between the sets of barriers and incentives/catalysts collectively using canonical correlation. RESULTS: Time costs, lack of technical support and large capital investments were the biggest barriers to computerization, whereas improved office efficiency and better-quality care were ranked highest as potential incentives to computerize. Cost vs. noncost, physician-related vs. patient-related, and monetary vs. nonmonetary factors were the key dimensions explaining the barrier variables. Similarly, within-practice vs external and "push" vs "pull" factors accounted for the incentive variables. Four clusters were identified for barriers and three for incentives/catalysts. Canonical correlation revealed that respondents who were concerned with the costs of computerization also perceived financial incentives and government regulation to be important incentives/catalysts toward computerization. Those who found the potential interference with communication important also believed that the promise of improved care from computerization to be a significant incentive. CONCLUSION: This study provided evidence regarding common barriers associated with clinical computerization. Our findings also identified possible incentive strategies that may be employed to accelerate uptake of computer systems.