By considering temporal properties including the study duration, we validate the consistency of the performance. In addition, the memory model is compared with conventional probabilistic model to supplier Taxol evaluate the performance of expectation in nonstationary environment. 3. Hypergraph-Based Memory Model We propose

a hypergraph-based memory model that enables incrementally encoding nonstationary contextual data and operating recognition judgment from the encoded memory model. In this section, we describe the memory mechanism, including encoding and judgment, from the concept of a hypergraph structure. The basic concept of the memory model follows the principles of a cognitive agent suggested by Zhang [34]. The hypergraph structure mimics brain mechanism related to memory encoding and retrieving. For memory encoding, input

data are disassembled into subsets and distributed for storage in memory. To retrieve the data, segmented subsets are composited to generate the complete data. The primary processes of memory encoding and judgment from the memory are partitioning and combining. To support these memory mechanisms based on a subset combination, we apply a hypergraph structure and modify the structure by constructing a layered hypergraphs. 3.1. Hypergraph-Based Memory Structure A hypergraph is a graphical model composed of edges, which are combinations of nodes [35]. When an event instance X is x1, x2,…, x6, a hypergraph can be represented as shown in Figure 1(a). In a hypergraph, a complete instance is divided into several subsets, which share a common property. Each node is allowed to be included in distinguished subsets according to the endowed parameter conditions. A single subset, combination of nodes, is assigned as a hyperedge with k nodes, where k is a variable indicating the size of the nodes in a subset. Figure 1 Graphical diagram of a hypergraph-based structure. (a) A hypergraph with six nodes and six edges. (b) A hypergraph structure constructs circular connections inside the network when the data comes from contextual events. The structure of a hypergraph has the advantage of building high-order relationships. Using the

flexible combinatorial structure of a hypergraph, several research domains have applied such characteristics as a spatial relationship in image processing and a temporal relationship in formal language analysis [36–38]. A hypergraph structure is adaptable to build relations of contextual Carfilzomib data and serial data. To make a dense connection inside the data, a hyperedge includes links with each weight between adjacent edges such that the hyperedges are fully connected. For example, if a hypergraph tries to model contextual event instances which are composed of six attributes, each edge is composed of k nodes including the node in the order of dimensions. In this case, the hypergraph structure is modified into the shape of a circular network, as shown in Figure 1(b).