Search

Encoding a dictionary into another representation means that all the words can be stored in the dictionary in a more efficient way. In this way, we can complete common operations in dictionaries, such as (1) searching for a word in the dictionary, (2) adding some words to the dictionary, and (3) removing some words from the dictionary, in a shorter time. Binary decision diagrams (BDDs) are one of the most famous representations of such encoding and are widely popular due to their excellent properties. Recently, some people have proposed encoding dictionaries into BDDs and some variants of BDDs and showed that it is feasible. Hence, we further investigate the topic of encoding dictionaries into decision diagrams. Tagged sentential decision diagrams (TSDDs), as one of these variants based on structured decomposition, exploit both the standard and zero-suppressed trimming rules. In this paper, we first introduce how to use Boolean functions to represent dictionary files and then design an algorithm that encodes dictionaries into TSDDs with the help of tries and a decoding algorithm that restores TSDDs to dictionaries. We utilize the help of tries in the encoding algorithm, which greatly accelerates the encoding process. Considering that TSDDs integrate two trimming rules, we believe that using TSDDs to represent dictionaries would be more effective, and the experiments also show this. [ABSTRACT FROM AUTHOR]

Abstract In the past decades, forgetting has been investigated for many logics and has found many applications in knowledge representation and reasoning. In this paper, we study forgetting in multi-agent modal logics. We adopt the semantic definition of existential bisimulation quantifiers as that of forgetting. We resort to canonical formulas of modal logics introduced by Moss. An arbitrary modal formula is equivalent to the disjunction of a unique set of satisfiable canonical formulas. We show that, for the logics of K n , D n , T n , K 45 n , KD 45 n and S 5 n , the result of forgetting an atom from a satisfiable canonical formula can be computed by simply substituting the literals of the atom with ⊤. Thus we show that these logics are closed under forgetting, and hence have uniform interpolation. Finally, we generalize the above results to include common knowledge of propositional formulas. [ABSTRACT FROM AUTHOR]

Knowledge compilation is an alternative solution to address demanding reasoning tasks with high complexity via converting knowledge bases into a suitable target language. The notion of logical separability, proposed by Levesque, offers a general explanation for the tractability of clausal entailment for two remarkable languages: decomposable negation normal form and prime implicates. It is interesting to explore what role logical separability plays in problem tractability. In this paper, we apply the notion of logical separability to a number of reasoning problems within the context of propositional logic: satisfiability checking (CO), clausal entailment checking (CE), model counting (CT), model enumeration (ME) and forgetting (FO), as well as their dual tasks, contributing to several recursive procedures. We provide the corresponding logical separability based properties: CO-logical separability, CE-logical separability, CT-logical separability, ME-logical separability and their duals. Based on these properties, we then identify four novel normal forms: CO - LSNNF , CE - LSNNF , CT - LSNNF and ME - LSNNF , as well as their dual languages. We show that each of them is the necessary and sufficient condition under which the corresponding procedure is correct. We finally integrate the above normal forms into the knowledge compilation map. [ABSTRACT FROM AUTHOR]

In this paper, we propose a novel reweighted price relative tracking (RPRT) system for automatic portfolio optimization (APO). In the price prediction stage, it automatically assigns separate weights to the price relative predictions according to each asset’s performance, and these weights will also be automatically updated. In the portfolio optimizing stage, a novel tracking system with a generalized increasing factor is proposed to maximize the future wealth of next period. Besides, an efficient algorithm is designed to solve the portfolio optimization objective, which is applicable to large-scale and time-limited situations. Extensive experiments on six benchmark datasets from real financial markets with diverse assets and different time spans are conducted. RPRT outperforms other state-of-the-art systems in cumulative wealth, mean excess return, annual percentage yield, and some typical risk metrics. Moreover, it can withstand considerable transaction costs and runs fast. It indicates that RPRT is an effective and efficient APO system. [ABSTRACT FROM AUTHOR]

Adversarial attacks hamper the functionality and accuracy of deep neural networks (DNNs) by meddling with subtle perturbations to their inputs. In this work, we propose a new mask-based adversarial defense scheme (MAD) for DNNs to mitigate the negative effect from adversarial attacks. Our method preprocesses multiple copies of a potential adversarial image by applying random masking, before the outputs of the DNN on all the randomly masked images are combined. As a result, the combined final output becomes more tolerant to minor perturbations on the original input. Compared with existing adversarial defense techniques, our method does not need any additional denoising structure or any change to a DNN's architectural design. We have tested this approach on a collection of DNN models for a variety of datasets, and the experimental results confirm that the proposed method can effectively improve the defense abilities of the DNNs against all of the tested adversarial attack methods. In certain scenarios, the DNN models trained with MAD can improve classification accuracy by as much as 90 % compared to the original models when given adversarial inputs. [ABSTRACT FROM AUTHOR]

The movement characteristics that nodes exhibit will bring serious problems in a Wireless Sensor Network (WSN) because most of the protocols do not adapt to node movement. This requires a mobility-aware evaluation technique to locate nodes, forecast communication quality, trace the movement trajectory and formulate thresholds to initiate future handover. One off-the-shelf technology using the Received Signal Strength Indicator (RSSI) of wireless radio is extensively adopted because of the small expense. A node can compute the relative distance between it and its partner by using RSSI readings. However, the measurement results of RSSI fluctuate significantly due to the effects of shadowing and fading. This paper is oriented to investigate the localization accuracy of RSSI in an outdoor ambient condition. To do this, some dynamic and stationary tests are carried out, a calibration line displaying the one-to-one match between RSSI and distance is mapped, five filtering approaches are developed to alleviate the fluctuation of movement curves, and the alleviation effect is measured by calculating the Root Mean Square Error (RMSE) values and by converting the time-domain filtering results into the Fast Fourier Transform (FFT) spectrum. Although the optimized RMSE is reduced to 0.86 and the noise FFT amplitude is less than 1 dBm, it may still have a single RSSI value corresponding to more than one distance. Moreover, these distances can differ as much as 2.8 m. Since the evaluation error is not acceptable for most cases, it is inaccurate for a mobile node to measure the distance from its partner rely on RSSI readings under outdoor scenarios. [ABSTRACT FROM AUTHOR]

Portfolio optimization (PO) has been catching more and more attention in the artificial intelligence and the machine learning communities. In this paper, we propose a novel Trend Representation based Log-density Regularization (TRLR) system for portfolio optimization. Its novelty falls into two aspects. First, it introduces a log-density regularization to the increasing factor of portfolio, which is seldom addressed by previous PO systems. It reflects a relationship between the portfolio and the price relative at an equilibrium point. Second, TRLR exploits a novel trend representation by taking the time variable as regressor in a weighted ridge regression, hence TRLR captures price trend patterns effectively. Extensive experiments conducted on 5 benchmark datasets from real-world financial markets demonstrate that TRLR achieves significantly better performance than other state-of-the-art strategies and runs fast, which shows its effectiveness and efficiency for large-scale applications. [ABSTRACT FROM AUTHOR]

• We propose to use a structural feature extraction scheme including 3 types of features (raw + tree decomposition + ECFP). • We design a framework including several graph attention convolutional (GAC) blocks and deep neural network (DNN) blocks to process the above structural features. • We design a readout block based on gated recurrent units (GRU). The readout blocks collaborate with the GAC blocks to obtain molecular embeddings. • We visualize molecules and mark the important atoms with attention scores of MSSGAT, which can be a good reference for subsequent drug development. Molecular machine learning based on graph neural network has a broad prospect in molecular property identification in drug discovery. Molecules contain many types of substructures that may affect their properties. However, conventional methods based on graph neural networks only consider the interaction information between nodes, which may lead to the oversmoothing problem in the multi-hop operations. These methods may not efficiently express the interacting information between molecular substructures. Hence, We develop a Molecular SubStructure Graph ATtention (MSSGAT) network to capture the interacting substructural information, which constructs a composite molecular representation with multi-substructural feature extraction and processes such features effectively with a nested convolution plus readout scheme. We evaluate the performance of our model on 13 benchmark data sets, in which 9 data sets are from the ChEMBL data base and 4 are the SIDER, BBBP, BACE, and HIV data sets. Extensive experimental results show that MSSGAT achieves the best results on most of the data sets compared with other state-of-the-art methods. [ABSTRACT FROM AUTHOR]