Perceptual organization in the human visual system involves neural mechanisms that spatially group and segment image areas based on local feature similarities, such as the temporal correlation of luminance changes. Successful segmentation models in computer vision, including graph-based algorithms and vision transformer, leverage similarity computations across all elements in an image, suggest that effective similarity-based grouping should rely on a global computational process. However, whether human vision employs a similarly global computation remains unclear due to the absence of appropriate methods for manipulating similarity matrices across multiple elements within a stimulus. To investigate how "temporal similarity structures" influence human visual segmentation, we developed a stimulus generation algorithm based on Vision Transformer. This algorithm independently controls within-area and cross-area similarities by adjusting the temporal correlation of luminance, color, and spatial phase attributes. To assess human segmentation performance with these generated texture stimuli, participants completed a temporal two-alternative forced-choice task, identifying which of two intervals contained a segmentable texture. The results showed that segmentation performance is significantly influenced by the configuration of both within- and cross-correlation across the elements, regardless of attribute type. Furthermore, human performance is closely aligned with predictions from a graph-based computational model, suggesting that human texture segmentation can be approximated by a global computational process that optimally integrates pairwise similarities across multiple elements.

Various acceleration approaches for Policy Gradient (PG) have been analyzed within the realm of Reinforcement Learning (RL). However, the theoretical understanding of the widely used momentum-based acceleration method on PG remains largely open. In response to this gap, we adapt the celebrated Nesterov's accelerated gradient (NAG) method to policy optimization in RL, termed *Accelerated Policy Gradient* (APG). To demonstrate the potential of APG in achieving fast convergence, we formally prove that with the true gradient and under the softmax policy parametrization, APG converges to an optimal policy at rates: (i) $\tilde{O}(1/t^2)$ with nearly constant step sizes; (ii) $O(e^{-ct})$ with time-varying step sizes. To the best of our knowledge, this is the first characterization of the convergence rates of NAG in the context of RL. Notably, our analysis relies on one interesting finding: Regardless of the parameter initialization, APG ends up entering a locally nearly-concave regime, where APG can significantly benefit from the momentum, within finite iterations. Through numerical validation and experiments on the Atari 2600 benchmarks, we confirm that APG exhibits a $\tilde{O}(1/t^2)$ rate with nearly constant step sizes and a linear convergence rate with time-varying step sizes, significantly improving convergence over the standard PG.

We revisit the domain of off-policy policy optimization in RL from the perspective of coordinate ascent. One commonly-used approach is to leverage the off-policy policy gradient to optimize a surrogate objective – the total discounted in expectation return of the target policy with respect to the state distribution of the behavior policy. However, this approach has been shown to suffer from the distribution mismatch issue, and therefore significant efforts are needed for correcting this mismatch either via state distribution correction or a counterfactual method. In this paper, we rethink off-policy learning via Coordinate Ascent Policy Optimization (CAPO), an off-policy actor-critic algorithm that decouples policy improvement from the state distribution of the behavior policy without using the policy gradient. This design obviates the need for distribution correction or importance sampling in the policy improvement step of off-policy policy gradient. We establish the global convergence of CAPO with general coordinate selection and then further quantify the convergence rates of several instances of CAPO with popular coordinate selection rules, including the cyclic and the randomized variants of CAPO. We then extend CAPO to neural policies for a more practical implementation. Through experiments, we demonstrate that CAPO provides a competitive approach to RL in practice.

Visual motion processing is essential for humans to perceive and interact with dynamic environments. Despite extensive research in cognitive neuroscience, image-computable models that can extract informative motion flow from natural scenes in a manner consistent with human visual processing have yet to be established. Meanwhile, recent advancements in computer vision (CV), propelled by deep learning, have led to significant progress in optical flow estimation, a task closely related to motion perception. Here we propose an image-computable model of human motion perception by bridging the gap between biological and CV models. Specifically, we introduce a novel two-stages approach that combines trainable motion energy sensing with a recurrent self-attention network for adaptive motion integration and segregation. This model architecture aims to capture the computations in V1-MT, the core structure for motion perception in the biological visual system, while providing the ability to derive informative motion flow for a wide range of stimuli, including complex natural scenes. In silico neurophysiology reveals that our model's unit responses are similar to mammalian neural recordings regarding motion pooling and speed tuning. The proposed model can also replicate human responses to a range of stimuli examined in past psychophysical studies. The experimental results on the Sintel benchmark demonstrate that our model predicts human responses better than the ground truth, whereas the state-of-the-art CV models show the opposite. Our study provides a computational architecture consistent with human visual motion processing, although the physiological correspondence may not be exact.

Nowadays, the number of comorbidities (physical-physical, mental-mental, physical-mental) is growing fast. The potential network structure of highly related diseases could be revealed and found by several approaches with the concept of disease network, such as Weighted Alternating Least Squares (WALS). The 2012 medical history of the Health Examination for the Elderly of Taipei City was used for this study. Based on traditional correlation analysis, the results show that physical and mental diseases/disorders have some special comorbid structure. Moreover, the correlations had some potential clusters with significant between-cluster separation. However, while we used WALS approach to explore the hidden structure of disease networks, the complex and unexpected disease networks of the aging population were revealed according to subjects' medical history. The hidden structure could be identified and further used for WALS calculating via matching controls who had no the specific disease to cases who carried it by other disease diagnosis. Our findings showed the predictive accuracy with 0.83 for the diagnostic model. It indicated the importance of hidden factors being used for further calculating the disease correlations of multisystem disorders.

The rapid growth in multimedia services and diverse mobile technologies have recently brought about sudden surges in traffic demands. To keep up with this trend, small cells are considered as a promising solution to supplement macrocell coverage and enhance network capacity in next generation cellular networks. However, the isotropic characteristics of conventional omni- directional antennas installed in small cells reveal a trade-off between coverage and power consumption. Boosted power levels allow for larger coverage areas while resulting in serious interference among cells. On the other hand, although weak power emissions can mitigate interference, coverage holes may appear instead. To overcome the physical limitations, a beamforming technique called switched multi- element antenna (SMEA) is introduced in this work, which is capable of shaping kinds of radiation patterns toward desired users. We propose an efficient radiation pattern selection scheme using SMEA, with the objective of maximizing the total number of served users within the coverage ranges of small cells. We then construct a series of simulations to evaluate the performance of our proposed scheme. Compared with a uniform power control scheme using omni-directional antenna and a distributed pattern selection scheme using SMEA, simulation results show that our proposed scheme can significantly enhance coverage and deal with the trade-off issue as well.

A rotation in a binary tree is a local restructuring that changes the tree into another and preserves the inorder sequence. The rotation distance between two binary trees is the minimum number of rotations needed to transform one into another. However, a polynomial-time algorithm for computing rotation distances between any two binary trees has still not been found. Lucas (The Computer Journal, 47, 259–269, 2004) recently presented an O(n2)-time algorithm for finding the exact rotation distance between two binary trees that are of a restricted form (each node has at most one child in the source tree and there is at most one zig-zag pair in the destination tree), where n is the number of nodes in each binary tree. In this paper, by using the coding technique of left weight sequences, which was proposed by Pallo (The Computer Journal, 9, 171–175, 1986), we find another restricted set of binary trees in which any two of them can be transformed with the exact rotation distance. Our algorithm can be performed in linear time for finding the rotation distance between any two trees in the restricted set. Moreover, the actual sequence of transforming rotations can also be built.

There are various ways to measure the shape difference between two n-node rooted binary trees (binary trees for short). A rotation on a binary tree is a local restructuring that changes the tree into another one preserving the in-order sequence. The rotation distance between two binary trees is the minimum number of rotations needed to transform one into another. Till now, no polynomial–time algorithm exists for computing the rotation distance between any two binary trees. Recently, Lucas (Comput. J., 47, 259–269, 2004) presented an O(n2)–time algorithm for finding the rotation distance between two binary trees, where the source tree is a degenerate tree and the destination tree is an angle tree. This paper improves the time-complexity to O(n) under this constraint.

One of the killer applications for radio frequency identification (RFID) system is the automatic identification of physical objects. To do that, each object must be labeled with a RFID tag containing identifying data. However, collisions occur when several tags send their data simultaneously. This paper proposes a secure scheme for zero-collision RFID tags identification based on code division multiple access (CDMA) and hash-chain mechanics. The scheme not only allows several tags send their identifying data simultaneously without collision, but also provides the following secure properties for RFID reader and tags: mutual authentication, tag anonymity, robustness and forward secrecy.