Automatic Construction of Fine-Grained Paraphrase Corpora System Using Language Inference Model

Paraphrase detection and generation are important natural language processing (NLP) tasks. Yet the term paraphrase is broad enough to include many fine-grained relations. This leads to different tolerance levels of semantic divergence in the positive paraphrase class among publicly available paraphrase datasets. Such variation can affect the generalisability of paraphrase classification models. It may also impact the predictability of paraphrase generation models. This paper presents a new model which can use few corpora of fine-grained paraphrase relations to construct automatically using language inference models. The fine-grained sentence level paraphrase relations are defined based on word and phrase level counterparts. We demonstrate that the fine-grained labels from our proposed system can make it possible to generate paraphrases at desirable semantic level. The new labels could also contribute to general sentence embedding techniques.

Rapid, Reference-Free Human Genotype Imputation with Denoising Autoencoders

Genotype imputation is a foundational tool for population genetics. Standard statistical imputation approaches rely on the co-location of large whole-genome sequencing-based reference panels, powerful computing environments, and potentially sensitive genetic study data. This results in computational resource and privacy-risk barriers to access to cutting-edge imputation techniques. Moreover, the accuracy of current statistical approaches is known to degrade in regions of low and complex linkage disequilibrium.Artificial neural network-based imputation approaches may overcome these limitations by encoding complex genotype relationships in easily portable inference models. Here we demonstrate an autoencoder-based approach for genotype imputation, using a large, commonly-used reference panel, and spanning the entirety of human chromosome 22. Our autoencoder-based genotype imputation strategy achieved superior imputation accuracy across the allele-frequency spectrum and across genomes of diverse ancestry, while delivering at least 4-fold faster inference run time relative to standard imputation tools.

History Marginalization Improves Forecasting in Variational Recurrent Neural Networks

Deep probabilistic time series forecasting models have become an integral part of machine learning. While several powerful generative models have been proposed, we provide evidence that their associated inference models are oftentimes too limited and cause the generative model to predict mode-averaged dynamics. Mode-averaging is problematic since many real-world sequences are highly multi-modal, and their averaged dynamics are unphysical (e.g., predicted taxi trajectories might run through buildings on the street map). To better capture multi-modality, we develop variational dynamic mixtures (VDM): a new variational family to infer sequential latent variables. The VDM approximate posterior at each time step is a mixture density network, whose parameters come from propagating multiple samples through a recurrent architecture. This results in an expressive multi-modal posterior approximation. In an empirical study, we show that VDM outperforms competing approaches on highly multi-modal datasets from different domains.

Spectral Probability Distribution of Closed Connected Water and Remote Sensing Statistical Inference for Yellow Substance

Unlike traditional remote sensing inversion, this study proposes a new distribution–distribution scheme, which uses statistical inferences to estimate the probability distribution of in-water components based on the probability distribution of the observed spectra. The distribution–distribution scheme has the advantages that it rapidly gives the statistical information of the water of interest, assists the traditional scheme in improving models, and provides more valuable information for water classification and aquatic environment analysis. In this study, based on Landsat-8 images, we analyzed the spectral probability distributions of 688 global waters and found that many of them were normal, log normal, and exponential distributions with diverse patterns in distribution parameters such as the mean, standard deviation, skewness, and kurtosis. Using simulated and field-measured data, we propose a bootstrap-based distribution–distribution scheme and develop some simple remote sensing statistical inference models to estimate the distribution parameters of yellow substance in water.

Feature blindness: a challenge for understanding and modelling visual object recognition

A central problem in vision sciences is to understand how humans recognise objects under novel viewing conditions. Recently, statistical inference models such as Convolutional Neural Networks (CNNs) seem to have reproduced this ability by incorporating some architectural constraints of biological vision systems into machine learning models. This has led to the proposal that, like CNNs, humans solve the problem of object recognition by performing a statistical inference over their observations. This hypothesis remains difficult to test as models and humans learn in vastly different environments. Accordingly, any differences in performance could be attributed to the training environment rather than reflect any fundamental difference between statistical inference models and human vision. To overcome these limitations, we conducted a series of experiments and simulations where humans and models had no prior experience with the stimuli. The stimuli contained multiple features that varied in the extent to which they predicted category membership. We observed that human participants frequently ignored features that were highly predictive and clearly visible. Instead, they learned to rely on global features such as colour or shape, even when these features were not the most predictive. When these features were absent they failed to learn the task entirely. By contrast, ideal inference models as well as CNNs always learned to categorise objects based on the most predictive feature. This was the case even when the CNN was pre-trained to have a shape-bias and the convolutional backbone was frozen. These results highlight a fundamental difference between statistical inference models and humans: while statistical inference models such as CNNs learn most diagnostic features with little regard for the computational cost of learning these features, humans are highly constrained by their limited cognitive capacities which results in a qualitatively different approach to object recognition.

Hybrid artificial intelligence-based inference models for accurately predicting dam body displacements: A case study of the Fei Tsui dam

Surveillance is a critical activity in monitoring the operation condition and safety of dams. This study reviewed the historical monitoring data of the Fei Tsui dam to determine possible influential factors for the dam body displacement and then evaluated the influencing degree of these factors by using correlation analysis. Thus, the key influential factors were identified objectively and further chosen as the input variables for numerous artificial intelligence (AI)-based inference models, including single machine learning techniques (support vector machine (SVM), artificial neural networks) and hybrid AI models. The models were trained and tested with 4722 real data retrieved in 11 years from the monitoring devices installed on elements of the dam, and then generated their respective inferred dam body displacement values. The results revealed that the adaptive time-dependent evolutionary least squares SVM model had the greatest performance by providing the lowest values of prediction errors in terms of mean absolute percentage error (MAPE = 8.14%), root mean square error (RMSE = 1.08 cm), and coefficient of determination (R = 0.993). The analysis results endorsed that the hybrid AI model could be an efficient tool to early produce accurate warnings of the dam displacements.

A survey on Deep Learning Based Eye Gaze Estimation Methods

In recent years, deep-learning systems have made great progress, particularly in the disciplines of computer vision and pattern recognition. Deep-learning technology can be used to enable inference models to do real-time object detection and recognition. Using deep-learning-based designs, eye tracking systems could determine the position of eyes or pupils, regardless of whether visible-light or near-infrared image sensors were utilized. For growing electronic vehicle systems, such as driver monitoring systems and new touch screens, accurate and successful eye gaze estimates are critical. In demanding, unregulated, low-power situations, such systems must operate efficiently and at a reasonable cost. A thorough examination of the different deep learning approaches is required to take into consideration all of the limitations and opportunities of eye gaze tracking. The goal of this research is to learn more about the history of eye gaze tracking, as well as how deep learning contributed to computer vision-based tracking. Finally, this research presents a generalized system model for deep learning-driven eye gaze direction diagnostics, as well as a comparison of several approaches.