A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is
Recent years have witnessed the great success of deep neural networks in many research areas. The fundamental idea behind the design of most neural networks is to learn similarity patterns from data for prediction and inference, which lacks the ability of logical reasoning. However, the concrete ability of logical reasoning is critical to many theoretical and practical problems. In this paper, we propose Neural Logic Network (NLN), which is a dynamic neural architecture that builds thearXiv:1910.08629v1 fatcat:634xfrpabrhshcl5k2irjo5tq4
more »... onal graph according to input logical expressions. It learns basic logical operations as neural modules, and conducts propositional logical reasoning through the network for inference. Experiments on simulated data show that NLN achieves significant performance on solving logical equations. Further experiments on real-world data show that NLN significantly outperforms state-of-the-art models on collaborative filtering and personalized recommendation tasks.
Recent years have witnessed the success of deep neural networks in many research areas. The fundamental idea behind the design of most neural networks is to learn similarity patterns from data for prediction and inference, which lacks the ability of cognitive reasoning. However, the concrete ability of reasoning is critical to many theoretical and practical problems. On the other hand, traditional symbolic reasoning methods do well in making logical inference, but they are mostly hardarXiv:2008.09514v1 fatcat:sbwv53hosneblezg6zodeifbzq
more »... reasoning, which limits their generalization ability to different tasks since difference tasks may require different rules. Both reasoning and generalization ability are important for prediction tasks such as recommender systems, where reasoning provides strong connection between user history and target items for accurate prediction, and generalization helps the model to draw a robust user portrait over noisy inputs. In this paper, we propose Logic-Integrated Neural Network (LINN) to integrate the power of deep learning and logic reasoning. LINN is a dynamic neural architecture that builds the computational graph according to input logical expressions. It learns basic logical operations such as AND, OR, NOT as neural modules, and conducts propositional logical reasoning through the network for inference. Experiments on theoretical task show that LINN achieves significant performance on solving logical equations and variables. Furthermore, we test our approach on the practical task of recommendation by formulating the task into a logical inference problem. Experiments show that LINN significantly outperforms state-of-the-art recommendation models in Top-K recommendation, which verifies the potential of LINN in practice.
The tree index structure is a traditional method for searching similar data in large datasets. It is based on the presupposition that most sub-trees are pruned in the searching process. As a result, the number of page accesses is reduced. However, time-series datasets generally have a very high dimensionality. Because of the so-called dimensionality curse, the pruning effectiveness is reduced in high dimensionality. Consequently, the tree index structure is not a suitable method for time-seriesdoi:10.1016/j.is.2004.05.001 fatcat:yn3jvkbqcnbg5gc5secentzjbe
more »... datasets. In this paper, we propose a two-phase (filtering and refinement) method for searching time-series datasets. In the filtering step, a quantizing time-series is used to construct a compact file which is scanned for filtering out irrelevant. A small set of candidates is translated to the second step for refinement. In this step, we introduce an effective index compression method named grid-based datawise dimensionality reduction (DRR) which attempts to preserve the characteristics of the time-series. An experimental comparison with existing techniques demonstrates the utility of our approach. r
Attract -PQLYSPECS is an acronym from Polymer NMR Spectral Database system and consists of H and 13C NMR spectra of commercial polymers all of which are originally obtained for the system. With the objective to help NMR specialists and polymer chemists elucidate molecular structures of organic polymers, 1H spectra were measured at 500 MHz and 125 MHz, respectively, for 206 different polymers, together with 13C at the corresponding frequencies. They were recorded with related information such asdoi:10.2116/analsci.7.supple_1601 fatcat:tyxply4ljfetxj23opzku6to3y
more »... experimental conditions as well as assignment of peaks in a CD-ROM of 540 Mbytes This system has functions to process spectra graphically, to analyze them quantitatively and to retrieve spectra via polymer names, structures, substructures, polymer classification codes, chemical shifts etc. and has been developed on a personal computer. The assignment were given to as many NMR signals as possible that correspond to main structures of homopolymers, as well as to copolymer sequences, stereoregular tacticities, branch structures, irregular bonds, end groups,and so on. The system has been designed to allow for users to develop their own elaborated functions such as estimation of chemical shifts and structural elucidation for advanced research of polymers. Although additivity rules for 13C chemical shifts are useful for those purpose, parameters in the rules are not always easy to obtain, especially for polymers which have long range interaction among subgroups in main chains and side chains. Parameters in additivity rules are easily decided by using appropriate sets of data in the databases. This is an example of sophisticated application of POLYSPECS, to which infrared spectra will be added in the next version to be issued soon. K r spectral database, polymer NMR, assignment, learning, PQLYSPECS, CD-ROM.
The laminar cooling of hot slabs is a distributed-parameter system characterized by time varying, nonlinear factors and physical difficulties in continuous temperature measurement. The process operation is classified into operating points according to steel No., thickness and target temperature. The moving slab is divided into segments upon which the control is to be implemented. The technology of soft sensing is incorporated into the temperature prediction model of the process that is adapteddoi:10.3182/20020721-6-es-1901.01566 fatcat:nusuelcw2vervkmlgxboe3epie
more »... nline to find the temperature profile through thickness, which will be used for online error correction in closed-loop control. Industrial experiments have shown the effectiveness of the proposed method. Copyright © 2002 IFAC
Collaborative Filtering (CF) has been an important approach to recommender systems. However, existing CF methods are mostly designed based on the idea of matching, i.e., by learning user and item embeddings from data using shallow or deep models, they try to capture the relevance patterns in data, so that a user embedding can be matched with appropriate item embeddings using designed or learned similarity functions. We argue that as a cognition rather than a perception intelligent task,arXiv:2005.08129v3 fatcat:x7plofpyl5drrcmhhnygpcwmmq
more »... dation requires not only the ability of pattern recognition and matching from data, but also the ability of logical reasoning in the data. Inspired by recent progress on neural-symbolic machine learning, we propose a neural collaborative reasoning framework to integrate the power of embedding learning and logical reasoning, where the embeddings capture similarity patterns in data from perceptual perspectives, and the logic facilitates cognitive reasoning for informed decision making. An important challenge, however, is to bridge differentiable neural networks and symbolic reasoning in a shared architecture for optimization and inference. To solve the problem, we propose a Modularized Logical Neural Network architecture, which learns basic logical operations such as AND, OR, and NOT as neural modules based on logical regularizer, and learns logic variables as vector embeddings. In this way, each logic expression can be equivalently organized as a neural network, so that logical reasoning and prediction can be conducted in a continuous space. Experiments on several real-world datasets verified the advantages of our framework compared with both traditional shallow and deep models.
Bovine herpesvirus1 (BoHV-1) is a major bovine pathogen. Despite several vaccines being available to prevent viral infection, outbreaks are frequent and cause important economic consequences worldwide. The development of new antiviral drugs is therefore highly desirable. In this context, viral genome replication represents a potential target for therapeutic intervention. BoHV-1 genome is a dsDNA molecule whose replication takes place in the nuclei of infected cells and is mediated by a viraldoi:10.3390/microorganisms8030409 pmid:32183205 pmcid:PMC7143239 fatcat:jrsnwrgkwraodfwp64wkmwojxy
more »... oded DNA polymerase holoenzyme. Here, we studied the physical interaction and subcellular localization of BoHV-1 DNA polymerase subunits in cells for the first time. By means of co-immunoprecipitation and confocal laser scanning microscopy (CLSM) experiments, we could show that the processivity factor of the DNA polymerase pUL42 is capable of being autonomously transported into the nucleus, whereas the catalytic subunit pUL30 is not. Accordingly, a putative classic NLS (cNLS) was identified on pUL42 but not on pUL30. Importantly, both proteins could interact in the absence of other viral proteins and their co-expression resulted in accumulation of UL30 to the cell nucleus. Treatment of cells with Ivermectin, an anti-parasitic drug which has been recently identified as an inhibitor of importin α/β-dependent nuclear transport, reduced UL42 nuclear import and specifically reduced BoHV-1 replication in a dose-dependent manner, while virus attachment and entry into cells were not affected. Therefore, this study provides a new option of antiviral therapy for BoHV-1 infection with Ivermectin.
Advances in Visual Information Management
Clustering is one of the most important topics in the field of knowledge discovery from databases. Specifically, hierarchical clustering is useful because it can be used to interactively guide users in browsing a huge database. In many cases, database clustering can be modeled as a graph partitioning problem, because a database with a distance function defined on it can be regarded as an edge weighted graph. So process of MST(Minimal Spanning Tree) construction is a possible solution to thisdoi:10.1007/978-0-387-35504-7_21 fatcat:ijnesvk4hfcahbleqerqgp44qi
more »... blem. In this paper, we propose an efficient MST construction method for a database with an arbitrary distance function on it. Our method utilizes a metric index to reduce the number of distance calculations needed to construct an MST. For this purpc;>se, we introduce a new metric index named metric matrix. Experimental results show that our method can reduce the number of distance calculations needed in comparison with the classical method.
Aggregate Nearest Neighbor Queries are much more complex than Nearest Neighbor queries, and pruning strategies are always utilized in ANN queries. Most of the pruning methods are based on the data index mechanisms, such as R-tree. But for the wellknown curse of dimensionality, ANN search could be meaningless in high dimensional spaces. In this paper, we propose two nonindex pruning strategies in ANN queries on metric space. Our methods utilize the r-NN query and projecting law, analyze thedoi:10.4108/infoscale.2007.900 dblp:conf/infoscale/LuoFCO07 fatcat:gvqtvmqwgrhoznxpthfjnzdt5a
more »... ibuting of query points, find out the search region in data space, and get the result efficiently.
In addition, besides high expression in gastric cancer, CacyBP/SIP is also highly expressed in pancreatic cancer , and plays a role in promoting cancer progression (Chen X. et al., 2011) . ...doi:10.3389/fmolb.2021.692941 pmid:34179100 pmcid:PMC8226165 fatcat:63yl7guvsvcx3n6uwunpg2itqy
Hanxiong Chen received the B.S. degree from Zhongshan University, Guangdong, China, in 1985, the M.S. degree and the Ph.D. degree in computer science, from the University of Tsukuba, Japan, in 1990 and ...doi:10.1587/transinf.2017dap0007 fatcat:iq6z2b52e5gihejeoqz4elb2xq
Advanced renal cancer is not sensitive to radiation and chemotherapy, and the efficacy of immunotherapy is very limited (Murai and Oya, 2004; Chen et al., 2015) . ... Copyright © 2021 Chen and Zheng. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). ...doi:10.3389/fmolb.2021.697962 pmid:34291088 pmcid:PMC8287069 fatcat:vo73xuik7za57bq4oee3vki2mi
2008 3rd International Conference on Intelligent System and Knowledge Engineering
Chen brought forward a method which can find RkNN for any k straightforwardly with constant running cost. ... Considering the cheap and large secondary storage and powerful computer doing update in spare time, Chen and etc argued that the pre-computing still finds its application when online costs, that is, CPU ...doi:10.1109/iske.2008.4730941 fatcat:5k66qj4swrgibh7dzfmhgj3guq
Author Biographies Hanxiong Chen received the B.S. degree from Zhongshan University, Guangdong, China, in 1985, the M.S. degree and the Ph.D degree in computer science, from the University of Tsukuba, ...doi:10.1007/s10115-010-0303-2 fatcat:yclhqlr3irf3bkzi6jt74jd6nu
The key approaches for machine learning, especially learning in unknown probabilistic environments are new representations and computation mechanisms. In this paper, a novel quantum reinforcement learning (QRL) method is proposed by combining quantum theory and reinforcement learning (RL). Inspired by the state superposition principle and quantum parallelism, a framework of value updating algorithm is introduced. The state (action) in traditional RL is identified as the eigen state (eigendoi:10.1109/tsmcb.2008.925743 pmid:18784007 fatcat:t6eb75ouynhtti5pdiibxphgxm
more »... ) in QRL. The state (action) set can be represented with a quantum superposition state and the eigen state (eigen action) can be obtained by randomly observing the simulated quantum state according to the collapse postulate of quantum measurement. The probability of the eigen action is determined by the probability amplitude, which is parallelly updated according to rewards. Some related characteristics of QRL such as convergence, optimality and balancing between exploration and exploitation are also analyzed, which shows that this approach makes a good tradeoff between exploration and exploitation using the probability amplitude and can speed up learning through the quantum parallelism. To evaluate the performance and practicability of QRL, several simulated experiments are given and the results demonstrate the effectiveness and superiority of QRL algorithm for some complex problems. The present work is also an effective exploration on the application of quantum computation to artificial intelligence.
« Previous Showing results 1 — 15 out of 127 results