As a key building block of biological cortex, neurons are powerful information processing units and can achieve highly complex nonlinear computations even in individual cells. Hardware implementation of artificial neurons with similar capability is of great significance for the construction of intelligent, neuromorphic systems. Here, we demonstrate an artificial neuron based on NbO x volatile memristor that not only realizes traditional all-or-nothing, threshold-driven spiking and spatiotemporal integration, but also enables dynamic logic including XOR function that is not linearly separable and multiplicative gain modulation among different dendritic inputs, therefore surpassing neuronal functions described by a simple point neuron model. A monolithically integrated 4 × 4 fully memristive neural network consisting of volatile NbO x memristor based neurons and nonvolatile TaO x memristor based synapses in a single crossbar array is experimentally demonstrated, showing capability in pattern recognition through online learning using a simplified δ-rule and coincidence detection, which paves the way for bio-inspired intelligent systems.

Optimization problems are ubiquitous in scientific research, engineering, and daily lives. However, solving a complex optimization problem often requires excessive computing resource and time and faces challenges in easily getting trapped into local optima. Here, we propose a memristive optimizer hardware based on a Hopfield network, which introduces transient chaos to simulated annealing in aid of jumping out of the local optima while ensuring convergence. A single memristor crossbar is used to store the weight parameters of a fully connected Hopfield network and adjust the network dynamics in situ. Furthermore, we harness the intrinsic nonlinearity of memristors within the crossbar to implement an efficient and simplified annealing process for the optimization. Solutions of continuous function optimizations on sphere function and Matyas function as well as combinatorial optimization on Max-cut problem are experimentally demonstrated, indicating great potential of the transiently chaotic memristive network in solving optimization problems in general.

To approach an advanced neuromorphic system, a significant unsettled problem is how to realize biologically plausible memory structures that are dramatically different from classical computers. Herein, a physical system based on memristors is simulated to realize associative memory based on discrete attractor networks, which is essentially content‐based storage, and the influence of device characteristics on network performance is systematically studied. An in situ unsupervised learning method is applied to make greater use of array structure and competitions between neurons, demonstrating significant performance improvement in memory capacity and noise tolerance compared with existing supervised approaches. By extending to continuous attractor neural networks (CANNs), working memory is realized based on memristors for the first time via simulation, and the write and read noises in memristor arrays are found to have different impacts on the ability of CANN in maintaining dynamic information. This work lays a foundation for the construction of future advanced neuromorphic computing systems.

Neuromorphic systems provide a potential solution for overcoming von Neumann bottleneck and realizing computing with low energy consumption and latency. However, the neuromorphic devices utilized to construct the neuromorphic systems always focus on artificial synapses and neurons, and neglected the important role of astrocyte cells. Here, an astrocyte memristor is demonstrated with encapsulated yttria‐stabilized zirconia (YSZ) to emulate the function of astrocyte cells in biology. Due to the high oxygen vacancy concentration and resultant high ionic conductivity of YSZ, significantly lower forming and set voltages are achieved in the artificial astrocyte, along with high endurance (>1011). More importantly, the nonlinearity in current‐voltage characteristics that usually emerge as the testing cycle increases can be depressed in the astrocyte memristor, and the nonlinearity can also be reversed by applying a refresh operation, which implements the role of biological astrocyte in maintaining the normal activity of neurons. The recovery of linearity can dramatically improve the accuracy of Modified National Institute of Standards and Technology dataset classification from 62.98% to 94.75% when the inputs are encoded in voltage amplitudes. The astrocyte memristor in this work with improved performance and linearity recovery characteristics can well emulate the function of astrocyte cells in biology and have great potential for neuromorphic computing.

Abstract. Matrix Factorization (MF) is one of the most popular approaches for recommender systems. Existing MF-based recommendation approaches mainly focus on the prediction of the users' ratings on unknown items. The performance is usually evaluated by the metric Root Mean Square Error (RMSE). However, achieving good performances in terms of RMSE does not guarantee a good performance in the top-N recommendation. Therefore, we advocate that treating the recommendation as a ranking problem. In this study, we present a ranking-oriented recommender algorithm AdaMF, which combines the MF model with AdaRank. Specifically, we propose an algorithm by adaptively combining component MF recommenders with boosting methods. The combination shows superiority in both ranking accuracy and model generalization. Normalized Discounted Cumulative Gain (NDCG) is chosen as the parameter of the coefficient function for each MF recommenders. In addition, we compare the proposed approach with the traditional MF approach and the state-of-the-art recommendation algorithms. The experimental results confirm that our proposed approach outperforms the state-of-the-art approaches.