Abstract: Real-time strategy (RTS) game has proposed many challenges for AI research for its large state spaces, enormous branch factors, limited decision time and dynamic adversarial environment. To tackle above problems, the method called Adversarial HierarchicalTaskNetwork planning (AHTN) has been proposed and achieves favorable performance. However, the HTN description it used cannot express complex relationships among tasks and impacts of environment on tasks. Moreover, the AHTN cannot handle task failures during plan execution. In this paper, we propose a modified AHTN planning algorithm named AHTNR. The algorithm introduces three elements essential task, phase and exit condition to extend the HTN description. To deal with possible task failures, the AHTNR first uses the extended HTN description to identify failed tasks. And then a novel task repair strategy is proposed based on historical information to maintain the validity of previous plan. Finally, empirical results are presented for the μRTS game, comparing AHTNR to the state-of-the-art search algorithms for RTS games.
Using partial domain models we have been able to replicate the GIPO-constructed operators and methods by induction as follows. Example files containing the partial domain (including an object class hierarchy) but excluding all the operators, methods and tasks are compiled. These files each contain a solution to a planning task in the form of a named operator sequence, initial states for the objects involved and numbered example material indicating the states after the application of each operator. The induction algorithm outputs a set of instantiated operators and an HTN method induced from the sequence. In each case only those operators required for the methods we were replicating were induced from each file. An example induced operator put_box_in_bag, where state transitions are written as pre-action state → post-action state, is as follows.
This paper proposes a new approach to extend the limitation of the previous algorithm by enabling their capability to generate a concurrent plan. The proposed algorithm based on HierarchicalTaskNetwork (HTN) enhances SHOP2 planning system to detect and generate a concurrent plan based on the output of SHOP2 (sequence plan). To trigger the availability of concurrent planning, allocation of resources based on web services inputs is used. The inputs (resources) are compared with SHOP2 operator, and concurrent plan will initialize if the instance of inputs equal to SHOP2 operator. To evaluate our approach, we perform two experiments using pathway information retrieval and logistic dataset from SHOP2 benchmark problem. The result of pathway information retrieval shows that this approach is able to find and generate a concurrent plan, but it takes longer computational time. Meanwhile, by using logistic dataset, the proposed algorithm is efficient to handle concurrent tasks based on cost reduction by some pruned operators. Therefore, in our future work, we intend to examine the approach in other complex Bioinformatics and system biology workflow which widely used web services as their analysis tools.
Wireless Sensor Networks (WSNs) consist of variable number of static or mobile nodes deployed across an area of interest to obtain, process and transmit relevant information. Wireless sensor node consist of several parts powered by battery , these includes: Sensing unit; this is capable of sensing temperature, humidity, visual, acoustic, location and many more. Micro-controller for processing the obtained data and a radio transceiver for transmitting the processed data to the Base Station (Sink) through a radio frequency channel. Sink node being a resourceful node having un-restricted communication, computational capability and additional energy source acts as an interface between wireless sensor networks and resource management center.The event being observed using WSNs may be static or dynamic depending on application scenario. Wireless Sensor Network have been used in many applications such as: Home automation and security, military applications for boarder control, intelligence and reconnaissance surveillance , environmental applications such as fire forest detection, health application for patient monitoring, agricultural usage, vehicle tracking, inventory management, civil aviation and a host of other applications. In addition to the numerous usage of WSNs, it equally provides a bridge between the physical and virtual world and allows the ability to observe the previously unobserved at a fine resolution over large spatio-temporal scales . Therefore, the relevance of this work to Aviation Industry cannot be over emphasized as WSNs is envisioned to effectively monitor trespass on restricted areas. Depending on application scene, wireless sensor node can be randomly distributed or uniformly placed .Wireless Sensor Networks are envisioned to operate in an autonomous and unrestricted fashion and its deployment has the potential to overcome the limitations associated with wired sensor networks, perimeter fencing and security patrol team as a way of monitoring trespass on Airport restricted areas.
The Genic Regulation Network (GRN) task consists of (1) extracting information on molecular interactions between genes and proteins that are described in scientific literature, and (2) using this information to reconstruct a regulation network between molecular partners in a formal way. Several other types of biological networks can be defined at the molecular level, such as metabolisms, gene expressions, protein- protein interactions or signaling pathways. All these networks are closely interconnected. For example, a gene codes for a protein that catalyzes the transformation of small molecules (metabolites), while the expression of the gene and its related regulation is controlled by other proteins.
In our current research work, we have proposed a solution to make use of the concept of survivability, based on active networking technology . This scheme enables MANETs to dynamically select parameters of MAC layer as well as network layer, and to dynamically select suitable protocols based on the requirements of an application as well as the communication environments. In order to meet the bandwidth requirements, the communicating nodes may change from the current MAC layer protocol and routing algorithm to a more capable routing algorithm. In case the environment of a cluster becomes harsher, cognitive networking enables the nodes to learn about the changes in the environment and take required actions for survivability. Robust networking with bandwidth protection strategy can avoid or efficiently deal with exception conditions occurred in the network. Figure 5 depicts our proposed conceptual framework, which integrates MAC layer and network layer together for preservation of bandwidth. The clustering algorithm and fault tolerance algorithm ensure the concept of survivability in the network. The QoS mechanisms ensure the guarantees to applications. Moreover, positioning scheme helps to optimize reservation of bandwidth.
round to the base station. The constrained load on the elected cluster heads during the 400 round of simulation drastically reduced the CHs’ energy over a short period. Unlike the non-hierarchical formation, the proposed hierarchical routing technique in which cluster hierarchy takes precedence in cluster formation and evaluate the residual energy for selection of cluster head, we observed that this technique offers a better life span for individual nodes and even the entire network. With optimization in energy usage, we observed that the lifetime in our proposed hierarchical technique extends to an impressive range when compared to non-hierarchical technique. The impressive increment in life span of the network from our proposed hierarchical technique is seen as a result of efficient routing decision and optimization of energy in cluster head selection of each cluster formed. Since the sensor nodes in each cluster send data to the cluster head within its cluster range and then the aggregated data is sent to the cluster head closer to the base station, which further aggregates data of its own cluster and that of the incoming data, from cluster head whose distance is farther to the BS, before sending the data to the base station. Thus, a considerable amount of energy is saved which indicate improved network lifetime in the case of first level hierarchy when compared to non hierarchical technique. From Fig. 6, we observed that the Non-hierarchical technique had an estimated lifetime of 20 rounds, First level had an estimated lifetime of 30 rounds and Second level had an estimated lifetime of 40 rounds. The progressive increase of network lifetime employed by our proposed technique offers efficient energy usage for each node in the entire network
LEACH (Low Energy Adaptive Cluster Hierarchy) ,  is a self- organizing, adaptive clustering protocol. It is based on TDMA based MAC protocol which is used to provide data aggregation with energy efficient communication. Each node follows stochastic (random) algorithm at each round to become CH. The information is transmitted from node to CH and from CH to BS. Aggregation works are carried out by CH. Random CH selection in each round with rotation. Within clusters, TDMA is preferred and CDMA is used across the clusters. To reduce collisions, inter cluster and intra clusters are used. LEACH assumes all the nodes having the ability to act as CH. The major advantage is, it organizes entire network distributed without any global knowledge, less power consumption because of aggregation by cluster heads. The disadvantages are that it assumes each and every node in the network with equal energy and transmits this data will cause battery drainage and all the nodes should adhere to both TDMA and CDMA techniques. Hierarchical cluster- based routing , clusters are organized only for a short span time termed as round. A round has two phases; election as well as data transfer phase. Here in election phase all the nodes are organized into different set clusters and these cluster heads consist of a headset. While in data transfer phase, the head set node will only involve in long range communication with the base station from cluster head. TEEN (Threshold sensitive energy efficient protocol)  is a reactive protocol because the nodes react immediately to sudden and drastic changes in the value of a sensed attribute. At every change of CH, that information can be broadcasted to its members. It is event driven protocol for time critical applications. There are two threshold levels
Word-level language model can only learn the relationship between words in one sentence. For sentences in one document which talks about one or several specific topics, the words in the next sentence are chosen partially in accordance with the previous sentences. To model this kind of co- herence of sentences, Le and Mikolov (2014) ex- tend word embedding learning network (Mikolov et al., 2013) to learn the paragraph embedding as a fixed-length vector representation for paragraph or sentence. Li and Hovy (2014) propose a neu- ral network coherence model which employs dis- tributed sentence representation and then predict the probability of whether a sequence of sentences is coherent or not.
Much effort has been devoted to evaluate whether multi-task learning can be leveraged to learn rich representations that can be used in various Natural Language Processing (NLP) down-stream applications. However, there is still a lack of understanding of the settings in which multi-task learning has a significant effect. In this work, we introduce a hierar- chical model trained in a multi-task learning setup on a set of carefully selected semantic tasks. The model is trained in a hierarchical fashion to introduce an inductive bias by su- pervising a set of low level tasks at the bottom layers of the model and more complex tasks at the top layers of the model. This model achieves state-of-the-art results on a number of tasks, namely Named Entity Recognition, Entity Mention De- tection and Relation Extraction without hand-engineered fea- tures or external NLP tools like syntactic parsers. The hierar- chical training supervision induces a set of shared semantic representations at lower layers of the model. We show that as we move from the bottom to the top layers of the model, the hidden states of the layers tend to represent more complex semantic information.
N-gram models have been used before (Goodman, 2001; Katz, 1987; Kneser and Ney, 1995). Re- cently, RNN-based models have achieved a better performance (Mikolov et al., 2010; J´ozefowicz et al., 2016; Grave et al., 2017; Melis et al., 2018). As conversational nature is not explicitly modeled, models often have role-switching issues. Task-Oriented Dialogue Systems. Conventional task-oriented dialog systems entails a sophisti- cated pipeline (Raux et al., 2005; Young et al., 2013) with components including spoken lan- guage understanding (Chen et al., 2016; Mesnil et al., 2015; Gupta et al., 2019), dialog state track- ing (Henderson et al., 2014; Mrksic et al., 2017), and dialog policy learning (Su et al., 2016; Gaˇsi´c and Young, 2014). Building a task-oriented dia- logue agent via end-to-end approaches has been explored recently (Li et al., 2017b; Wen et al., 2017). Although several conversational datasets are published recently (Gopalakrishnan et al., 2019; Henderson et al., 2019), the scarcity of an- notated conversational data remains a key problem when developing a dialog system. This motivates us to model task-oriented dialogues with goal in- formation in order to achieve controlled dialogue generation for data augmentation.
Language models are a major class of natural lan- guage processing (NLP) models whose develop- ment has lead to major progress in many areas like translation, speech recognition or summarization (Schwenk, 2012; Arisoy et al., 2012; Rush et al., 2015; Nallapati et al., 2016). Recently, the task of language modeling has been shown to be an ad- equate proxy for learning unsupervised represen- tations of high-quality in tasks like text classifica- tion (Howard and Ruder, 2018), sentiment detec- tion (Radford et al., 2017) or word vector learning (Peters et al., 2018).
Our translation system belongs to the hierarchi- cal phrase-based class (Chiang, 2007), i.e. phrase pairs with nonterminals (rules of a synchronous context-free grammar) are extracted from sym- metrized word alignments and subsequently used by the decoder. We use Joshua, a Java-based open- source implementation of the hierarchical decoder (Li et al., 2009), release 1.3. 1
Bongale et al. (2016) pointed out that after the middleware technology was introduced into the wireless sensor network, many research institutions began to study from different aspects. The vast majority of middleware development efforts focus on extending the life cycle of the network and how to maximize the use of the network's limited resources . In addition, Solaiman et al. (2016) also pointed out that many famous universities in the United States, such as the University of California at Berkeley, have begun to explore the theoretical basis and engineering application technology of wireless sensor networks . At the same time, the military departments and industries of many countries have also turned their attention to this field. Arya et al. (2016) pointed out in the study that the Natural Science Foundation of the United States established and implemented a research plan for wireless sensor networks in 2003; military units, including the US Department of Defense, have given considerable attention to wireless sensor networks and established related research projects . Elhoseny et al. (2016) pointed out that the famous companies in the information industry such as Intel and Microsoft in the United States have also carried out research work in this area. The developed countries such as Japan, Germany, the United Kingdom, and Italy have also started research in this field .
A number of processors in bus based System on Chip (SoC) are increased continuously and they face design challenges in different aspects . This bus architecture has faced bottleneck problem when more processors in- tegrated into single chip. To avoid bottleneck, bus archi- tecture is replaced with network architecture which is similar to the data networks. This new technology is known as Network on Chip (NoC) and it is widely ac- cepted as a solution for communication issues in SoC. Data communication between the processors is pack- etized and transmitted throughout the entire network [2, 3]. The basic components of NoC are processors, memo- ries, routers and physical links. All the processors, mem- ory blocks and other cores are connected to routers using physical links. The routers are interconnected to each other directly or through other intermediate routers. The role of router is to make decision where the data is to be transmitted based on destination address in the header flit of message packet [4-6]. A routing algorithm plays a ma- jor role in NoC that helps to communicate one processor to other processors or memory. This paper presents dif- ferent routing algorithms such as XY—routing algo- rithm, OE—turn routing algorithm, and Pseudo adaptive routing algorithm. Additionally new algorithm has also proposed in this paper to achieve better performance for different NoC architectures. These routing algorithms
The transition between descending, then crossing, and back to descending gas desorption scanning curves observed for the silica-alumina catalyst is also consistent with the pore model shown on Figure 12. The filling process for the highly-simplified, model network shown in Figure 12 would commence with capillary condensation via the cylindrical sleeve- shaped meniscus within the smallest through pores. If at the ultimate highest pressure of the scanning curve, the smallest through pores were filled with liquid condensate but the other larger pores were left empty, then desorption from the smallest pores would commence from the hemispherical menisci at the junctions of the smallest pores and the other larger pores. According to the Cohan equations, this would result in wide hysteresis, and, thence, a crossing scanning curve. This is what is observed for scanning curves turning around at the kink at relative pressure of 0.66 in the boundary adsorption isotherm. Once the smallest pores in the model are full of condensate, the medium pores can potentially fill by advanced
Abstract. Medical image segmentation is a basic step in medical image analysis, especially for medical image sequences such as CT sequences. Automated segmentation of different objects in the medical image sequences is of great significance to the 3D reconstruction of medical images. A novel image recognition method which can be implemented in automated medical image segmentation is introduced. In contrast with other algorithm, HTM (hierarchical temporal memory) is a network using a spatio-temporary hierarchy that works as our neocortex. The algorithm refereed in this paper consists of three main steps. Firstly, a four level hierarchical structure is established. Secondly, create frames by animating gray images to train the HTM network. During the learning phase, the nodes in HTM network build its representations spatial pooler and temporal pooler for inputs. Thirdly, test with dataset to get the inference result for classification. The results show that the proposed method can recognize the “middle slice” for different given objects when process the medical image sequences.
A long user history inevitably reflects the transitions of per- sonal interests over time. The analyses on the user history require the robust sequential model to anticipate the tran- sitions and the decays of user interests. The user history is often modeled by various RNN structures, but the RNN structures in the recommendation system still suffer from the long-term dependency and the interest drifts. To resolve these challenges, we suggest HCRNN with three hierarchical con- texts of the global, the local, and the temporary interests. This structure is designed to withhold the global long-term interest of users, to reflect the local sub-sequence interests, and to attend the temporary interests of each transition. Be- sides, we propose a hierarchical context-based gate structure to incorporate our interest drift assumption. As we suggest a new RNN structure, we support HCRNN with a comple- mentary bi-channel attention structure to utilize hierarchical context. We experimented the suggested structure on the se- quential recommendation tasks with CiteULike, MovieLens, and LastFM, and our model showed the best performances in the sequential recommendations.
Abstract— Wireless Sensor Network (WSN) is one of the major research areas in computer network field today. The function of WSN in this paper is to provide sensing services in an un-attended harsh environment. Sensed data need to be delivered to the sink and to cope with the network unreliability problem. Few routing protocol takes into consideration of this problem. It is a great challenge of the hierarchical routing protocol to provide network survivability and redundancy features. In this paper, a short literature review of the existing routing protocol is carried out. Then a novel hierarchical routing protocol, which addresses network survivability and redundancy issues, is introduced. Initial analysis shows promising results of the proposed protocol over LEACH. Finally, conclusion was drawn based on the research and future direction for further research is identified.
In wireless sensor networks (WSNs), data transmission is secured by authenticating secret keys. Secure key management is most important for network reliability and consistency. In this paper, a hierarchical group key management technique using threshold cryptography in Wireless Sensor Networks is proposed. The technique considers hierarchical sensor network, where sensing nodes are coordinated by forwarding nodes (FN) and in turn they are connected to the BS which is responsible for key computation and distribution. FN estimates the group key using threshold secret sharing scheme. The acquired group key is divided into multiple shares and shared among member nodes. Thus, this reduces the possibility of key compromised. The proposed technique is simulated using network simulator 2 (NS-2). Simulation results show the proficiency of the technique .