Sohu (China)

Adversarial waveform generation has been a popular approach as the backend of singing voice conversion (SVC) to generate high-quality singing audio. However, the instability of GAN also leads to other problems, such as pitch jitters and U/V errors. It affects the smoothness and continuity of harmonics, hence degrades the conversion quality seriously. This paper proposes to feed harmonic signals to the SVC model in advance to enhance audio generation. We extract the sine excitation from the pitch, and filter it with a linear time-varying (LTV) filter estimated by a neural network. Both these two harmonic signals are adopted as the inputs to generate the singing waveform. In our experiments, two mainstream models, MelGAN and ParallelWaveGAN, are investigated to validate the effectiveness of the proposed approach. We conduct a MOS test on clean and noisy test sets. The result shows that both signals significantly improve SVC in fidelity and timbre similarity. Besides, the case analysis further validates that this method enhances the smoothness and continuity of harmonics in the generated audio, and the filtered excitation better matches the target audio.

News recommendation plays an indispensable role in acquiring daily news for users. Previous studies make great efforts to model high-order feature interactions between users and items, where various neural models are applied (e.g., RNN, GNN). However, we find that seldom efforts are made to get better representations for news. Most previous methods simply adopt pre-trained word embeddings to represent news and also suffer from cold-start users. In this work, we propose a new textual content representation method by building a word graph for recommendation, which is named WG4Rec. Three types of word associations are adopted in WG4Rec for content representation and user preference modeling, namely: 1)semantically-similar according to pre-trained word vectors, 2)co-occurrence in documents, and 3)co-click by users across documents. As extra information can be unified by adding nodes/edges to the word graph easily, WG4Rec is flexible to make use of cross-platform and cross-domain context for recommendation to alleviate the cold-start issue. To the best of our knowledge, it is the first attempt that using these relationships for news recommendation to better model textual content and adopt cross-platform information. Experimental results on two large-scale real-world datasets show that WG4Rec significantly outperforms state-of-the-art algorithms, especially for cold users in the online environment. Besides, WG4Rec achieves better performances when cross-platform information is utilized.

Knowledge graph (KG) embedding aims to encode entities and relations into low-dimensional vector spaces, in turn, can support various machine learning models on KG related tasks with good performance. However, existing methods for knowledge graph embedding fail to consider the influence of the embedding space, which makes them still unsatisfactory in practical applications. In this study, we try to improve the expressiveness of the embedding space from the perspective of the metric. Specifically, we first point out the implications of Minkowski metric used in KG embedding and then make a quantitative analysis. To solve the limitations, we introduce a new metric, named Cycle metric, based on the oscillation property of the periodic function. Furthermore, we find that the function period has a significant influence on the expressiveness of the embedding space. Given a fully trained model, the smaller the period, the better the expressive ability. Finally, to validate the findings, we propose a new model, named CyclE by combining Cycle Metric and the popular KG embeddings models. Comprehensive experimental results show that Cycle is more appropriate than Minkowski for KG embedding.

Identification of significant biological relationships or patterns is central to many metagenomic studies. Methods that estimate association networks have been proposed for this purpose; however, they assume that associations are static, neglecting the fact that relationships in a microbial ecosystem may vary with changes in environmental factors (EFs), which can result in inaccurate estimations. Therefore, in this study, we propose a computational model, called the k-Lognormal-Dirichlet-Multinomial (kLDM) model, which estimates multiple association networks that correspond to specific environmental conditions, and simultaneously infers microbe–microbe and EF–microbe associations for each network. The effectiveness of the kLDM model was demonstrated on synthetic data, a colorectal cancer (CRC) dataset, the Tara Oceans dataset, and the American Gut Project dataset. The results revealed that the widely-used Spearman’s rank correlation coefficient method performed much worse than the other methods, indicating the importance of separating samples by environmental conditions. Cancer fecal samples were then compared with cancer-free samples, and the estimation achieved by kLDM exhibited fewer associations among microbes but stronger associations between specific bacteria, especially five CRC-associated operational taxonomic units, indicating gut microbe translocation in cancer patients. Some EF-dependent associations were then found within a marine eukaryotic community. Finally, the gut microbial heterogeneity of inflammatory bowel disease patients was detected. These results demonstrate that kLDM can elucidate the complex associations within microbial ecosystems. The kLDM program, R, and Python scripts, together with all experimental datasets, are accessible at https://github.com/tinglab/kLDM.git.

Commonsense understanding is a long-term goal of natural language processing yet to be resolved. One standard testbed for commonsense understanding isStory Cloze Test (SCT) \citemostafazadeh2016corpus, In SCT, given a 4-sentences story, we are expected to select the proper ending out of two proposed candidates. The training set in SCT only contains unlabeled stories, previous works usually adopt the small labeled development data for training, which ignored the sufficient training data and, essentially, not reveal the commonsense reasoning procedure. In this paper, we propose an unsupervised sequence-to-sequence method for story reading comprehension, we only adopt the unlabeled story and directly model the context-target inference probability. We propose a loss-reweight training strategy for the seq-to-seq model to dynamically tuning the training process. Experimental results demonstrate the advantage of the proposed model and it achieves the comparable results with supervised methods on SCT.

Web collection is essential for many Web based researches such as Web Information Retrieval (IR), Web data mining, Corpus linguistics and so on. However, it is usually expensive and time-consuming to collect a large scale of Web pages in lab-based environment and public-available collection becomes a necessity for these researches. In this study, we present a Chinese Web collection, SogouT-16, which is the largest free-of-charge public Chinese Web collection so far. We provide a variety of descriptive characteristics of SogouT-16 and discuss its adoption in a newly-designed ad-hoc retrieval task in NTCIR-13, We Want Web. SogouT-16 also provides online retrieval service and contains a number of auxiliary resources including hyperlink structure graph, query logs, word embedding, and etc. We believe that SogouT-16 will provide new opportunities for novel investigations and applications in IR and other related communities.

Supply chain has been one of the most popular and efficient ways of exploring more benefits across businesses from the perspective of particular enterprise, with the objective of improving the long-term performance of individual companies. With the rapid development of social media platforms, such as Twitter, Facebook, LinkedIn, etc., social media has become one of the most valuable businesses due to largest user coverage, widest influential dissemination and highest commercial value. From the perspective of enterprises, leveraging social media for e-commerce are gradually being taken seriously. Indeed, social media helps compress the supply chain by eliminating unnecessary intermediate links. In this paper, we consider the roles in Twitter users (i.e., brands and audiences) as the nodes in supply chain, and simulate the social interactions between them as the collaboration decisions among supply chain members. Specifically, we consider social interactions as transactions, and sentiments as trust information. In this way, we tailor the traditional game model for modeling relationships among supply chain members in the context of Twitter, and propose a long-term game model based on trust.

In this paper, the problem of delay-dependent absolute stability condition for a new class of Takagi-Sugeno (T-S) fuzzy Lurie control systems with time-varying delay is investigated. New Lyapunov functions are defined by utilizing the Lyapunov-Krasovskii functional (LKF). Meanwhile, free-weighting matrix method and some techniques are used to obtain a new absolute stability condition, which is different from previous research. Finally, a numerical example will be provided to demonstrate feasibility and less conservativeness of the proposed result.

Alloy 690TT is a widely used material for steam generator tubes in pressurized water reactors, where it is susceptible to corrosion fatigue. In this study, the effect of loading frequency on the corrosion fatigue behavior of Alloy 690TT in boron-lithium water was investigated under small-scale yielding conditions. Along with stress ratio, the loading frequency had a significant influence on the crack path and fatigue crack growth rate. Both the corrosion fatigue crack growth rate and the environment enhancement factor were quantitatively correlated with the loading frequency in the form of exponential functions.

Our objective was to compare general and epidural anaesthesia for surgical resection of giant intraperitoneal tumours. 80 patients with giant intraperitoneal tumours were randomly divided into two groups which respectively received general or epidural anaesthesia prior to the surgery. General anaesthesia was done under intubation and mechanical ventilation. Vital signs, including diastolic pressure, systolic pressure, heart rate and oxygen saturation, were not significantly different between two groups before the surgery. Vital signs were decreased in patients of both groups after surgery. However, vital signs in the general anaesthesia group were markedly higher than in the epidural anaesthesia group (p < 0.05), indicating that the changes of vital signs were less pronounced in the general anaesthesia group. Furthermore, the operation time in this group was markedly shorter than in the epidural anaesthesia group, while the success rate was markedly higher. In addition, the intraoperative blood loss and infusion volume were both markedly lower in the general anaesthesia group, as well as the occurrence rate of supine position (p < 0.05 for all comparisons). Occurrence of heart failure, lung edema, altered water-electrolyte balance, ECG abnormalities in patients undergoing surgery under general anaesthesia group was significantly lower (p < 0.05 vs. patients under epidural anaesthesia). There were no deaths in either patient group, and all patients were successfully cured and discharged. General anaesthesia under intubation and mechanical ventilation is better suited for surgical removal of giant intraperitoneal tumours.