Search

Deep Reinforcement Learning (DRL) is one powerful tool for varied control automation problems. Performances of DRL highly depend on the accuracy of value estimation for states from environments. However, the Value Estimation Network (VEN) in DRL can be easily influenced by the phenomenon of catastrophic interference from environments and training. In this paper, we propose a Dynamic Sparse Coding-based (DSC) VEN model to obtain precise sparse representations for accurate value prediction and sparse parameters for efficient training, which is not only applicable in Q-learning structured discrete-action DRL but also in actor–critic structured continuous-action DRL. In detail, to alleviate interference in VEN, we propose to employ DSC to learn sparse representations for accurate value estimation with dynamic gradients beyond the conventional ℓ 1 norm that provides same-value gradients. To avoid influences from redundant parameters, we employ DSC to prune weights with dynamic thresholds more efficiently than static thresholds like ℓ 1 norm. Experiments demonstrate that the proposed algorithms with dynamic sparse coding can obtain higher control performances than existing benchmark DRL algorithms in both discrete-action and continuous-action environments, e.g., over 25% increase in Puddle World and about 10% increase in Hopper. Moreover, the proposed algorithm can reach convergence efficiently with fewer episodes in different environments. [ABSTRACT FROM AUTHOR]

Impact of covalent and non-covalent binding interactions on physicochemical, structural, and rheological properties of waxy rice flour were investigated using several typical types of chemicals. Dynamic rheometer was used to investigate the viscoelastic properties and resistance to deformation or shearing. Both hydrogen bonds and hydrophobic interactions were necessary for waxy rice flour to form a material with stable structure matrix. Besides this, X-ray diffraction pattern and pasting property measurements showed that the relative crystallinity and pasting viscosity values of waxy rice flour exposed to SDS and l -cysteine treatments increased whereas those of urea decreased compared with the control. Addition of acid, alkali, and NaCl might result in electrostatic forces, ion interactions, and indirectly change hydrogen bonds and hydrophobic interactions between starch and protein, which played a synthetic role on property changes of waxy rice flour. [ABSTRACT FROM AUTHOR]

Recently there has been increasing attention towards analysis dictionary learning. In analysis dictionary learning, it is an open problem to obtain the strong sparsity-promoting solutions efficiently while simultaneously avoiding the trivial solutions of the dictionary. In this paper, to obtain the strong sparsity-promoting solutions, we employ the ℓ 1 ∕ 2 norm as a regularizer. The very recent study on ℓ 1 ∕ 2 norm regularization theory in compressive sensing shows that its solutions can give sparser results than using the ℓ 1 norm. We transform a complex nonconvex optimization into a number of one-dimensional minimization problems. Then the closed-form solutions can be obtained efficiently. To avoid trivial solutions, we apply manifold optimization to update the dictionary directly on the manifold satisfying the orthonormality constraint, so that the dictionary can avoid the trivial solutions well while simultaneously capturing the intrinsic properties of the dictionary. The experiments with synthetic and real-world data verify that the proposed algorithm for analysis dictionary learning can not only obtain strong sparsity-promoting solutions efficiently, but also learn more accurate dictionary in terms of dictionary recovery and image processing than the state-of-the-art algorithms. [ABSTRACT FROM AUTHOR]

In this study, we propose two analysis dictionary learning algorithms for sparse representation with analysis model. The problem is formulated with the ℓ 1 -norm regularizer and with two penalty terms on the analysis dictionary: the term of − log det ( Ω T Ω ) and the coherence penalty term. As the processing scheme, we employ a block coordinate descent framework, so that the overall problem is transformed into a set of minimizations of univariate subproblems with respect to a single-vector variable. Each subproblem is still nonsmooth, but it can be solved by a proximal operator and then the closed-form solutions can be obtained directly and explicitly. In particular, the coherence penalty, excluding excessively similar or repeated dictionary atoms, is solved at the same time as the dictionary update, thereby reducing the complexity. Furthermore, a scheme with a group of atoms is introduced in one proposed algorithm, which has a lower complexity. According to our analysis and simulation study, the main advantages of the proposed algorithms are their greater dictionary recovery ratios especially in the low-cosparsity case, and their faster running time of reaching the stable values of the dictionary recovery ratios and the recovery cosparsity compared with state-of-the-art algorithms. In addition, one proposed algorithm performs well in image denoising and in noise cancellation. [ABSTRACT FROM AUTHOR]

[Display omitted] • A method for rapid and accurate detection of soil organic matter content. • The method is based on reflectance spectroscopy and machine learning. • Quality band combinations are obtained by spectral pre-processing methods. • Genetic strategy is introduced into the model to enhance its exploration capability. • We propose an improved hybrid extreme learning machine model. The continuous exploitation of mining resources and a number of other factors have led to serious soil pollution problems. Soil organic matter (SOM) can be used as an important measure of soil fertility and ecological environment quality in mining areas. The traditional method of measuring SOM content by chemical methods is costly and time-consuming, so it is important to develop a rapid, accurate and economical method to measure SOM content in reclaimed mines for the ecological restoration and the evaluation of the ecological reclamation effect of mines. In this paper, the visible-near infrared spectral data of soil were obtained using an open pit mine as the study area. The three pre-processing methods of multiple scattering correction, continuous removal and Savitzky-Golay smoothing were compared. Then the two-band ratio spectral index RI was constructed to obtain the sensitive band combinations. To address the problem that the HHO-TELM hybrid model tended to converge prematurely, the GHO-TELM model was proposed in this paper. GHO-TELM introduced a gene learning strategy in HHO-TELM. The crossover, mutation and selection mechanisms were used to cultivate high-quality examples, enhance the quality of individuals and prevent premature convergence of the model. GHO-TELM was compared with several machine learning models. The experimental results showed that GHO-TELM obtained the best performance indexes, with correlation coefficient (r) of 0.9586, and relative percentage deviation (RPD) of 3.0823, indicating that the model had a good prediction effect. [ABSTRACT FROM AUTHOR]

Mussel-inspired chemistry (polydopamine) offers great opportunities to develop inexpensive and efficient process for many types of materials with complex shapes and functions in a mild and friendly environment. This paper describes a facile, yet green approach to synthesize polydopamine/silver (PDA/Ag) nanocomposite particles with a combination use of polydopamine chemistry and electroless metallization of Ag. In this approach, monodisperse spherical polydopamine particles are first synthesized by the oxidation and self-polymerization of dopamine (monomer) in an alkaline water–ethanol solution at room temperature, which are served as the active templates for secondary reactions due to the abundant catechol and amine groups on the surface. Subsequently, the silver precursor-[Ag(NH 3 ) 2 ] + ions introduced are easily absorbed onto the surface of the PDA particles, and are immediately in situ reduced to metallic Ag nanoparticles with the help of these active catechol and amine groups. During the preparation, no additional reductants, toxic reagents and intricate instruments are needed. These as-synthesized PDA/Ag nanocomposite particles are ideal candidates for antibacterial application because they do not show significant cytotoxicity against HEK293T human embryonic kidney cells in the in vitro cytotoxicity assay, whereas demonstrate enhanced antibacterial abilities against Escherichia coli (Gram-negative bacteria) and Staphylococcus aureus (Gram-positive bacteria) in the antibacterial assays. Owing to their excellent cytocompatibilities and antibacterial activities, these PDA/Ag nanocomposite particles can be considered as the promising antibacterial materials for future biomedical applications. [ABSTRACT FROM AUTHOR]