Catálogo de publicaciones - libros

This paper proposes an improved voice activity detection (VAD) algorithm for controlling discontinuous transmission (DTX) of the GSM adaptive multi-rate (AMR) speech codec. First, based on the wavelet transform, the original IIR filter bank and the open-loop pitch detector are implemented via the wavelet filter bank and the wavelet-based pitch detection algorithm, respectively. The proposed wavelet filter bank divides the input speech signal into 9 frequency bands so that the signal level at each sub-band can be calculated. In addition, the background noise can be estimated in each sub-band by using the wavelet de-noising method. The wavelet filter bank is also derived to detect correlated complex signals like music. Then one can apply support vector machine (SVM) to train an optimized non-linear VAD decision rule involving the sub-band power, noise level, pitch period, tone flag, and complex signals warning flag of input speech signals. By the use of the trained SVM, the proposed VAD algorithm can produce more accurate detection results. Various experimental results carried out from the Aurora speech database show that the proposed algorithm gives considerable VAD performances superior to the AMR VAD Option 1 and comparable with the AMR VAD Option 2.

In this paper, the performance influencing class analysis (PI-CA) framework is proposed for performance analysis of pattern recognition systems dealing with data with great variety and diversity. Through the PICA procedure, the population of data is divided into subsets on which the system achieves different performances by means of statistical methods. On basis of the division, performance assessment and analysis are conducted to estimate the system performance on the whole data population. The PICA framework can predict true performance in real application and facilitate comparison of different systems without the same test set. The PICA framework is applied to the analysis of a broadcast news segmentation system. The procedure is presented and experimental results were given, which verified the effectiveness of PICA.

This paper proposes a new theory for forming a maximum-value-cooperation coalition known as . This theory can give such solution because it does not assume that each player requesting to join a coalition knows information of other players. However, all players’ private information requesting to join the coalition is known by an honest coordinator. This allows the coordinator to select a coalition structure with maximal value of cooperation among successful players so as they get at least at their required minimum values. Not only this maximal value is shown to be equal to or larger than the value of a core coalition but also the value allocation is shown to be Pareto optimal.

Conventional collaborative recommendation approaches neglect weak relationships even when they provide important information. This study applies the concepts of chance discovery and small worlds to recommendation systems. The trust (direct or indirect) relationships and product relationships among customers are to find candidates for collaboration. The purchasing quantities and feedback of customers are considered. The whole similarities are calculated based on the model, brand and type of purchased product.

In this paper, we handle a negotiation method in which a main negotiation consists of multiple sub-negotiations. In items allocation for commerce, there are some risks on the trade because the market balance is determined by supply and demand. The result of main negotiation is also determined by the order of sub-negotiations and agents’ behaviors since agents’ budgets have limitations on the actual commercial trading. However, it is difficult to decide the order of negotiations, such as simultaneous decisions and rotations. In this paper, we give the trading model in such cases, that is, agents purchase items by pooled buying. In actual trading as pooled buying, items are sold by the volume discount. Concretely, we discuss joint-stock company and private limited partnership on the web. In the negotiation phase, an agent proposes pooled buying based on the number of items and their prices considering their budgets. The degree of risks is calculated. All agents can see the risks with each item. Agents cooperate to the proposing agent based on the degree of risks. In this paper, we give two scenarios for trading. One is to avoid free riders who get surplus without risks. Another one is to promote agents’ participation to increase social surplus. For risk aversion and promoting cooperation, we employ the side-payment policy, that is, cooperative agents’ risks are preserved to a minimum. Further, we give some discussions where agents must pay the negotiation costs and charge for storage.

It is widely acknowledged that a temporal representation of e-contracts is essential in order to support e-contract execution and performance monitoring. One possibility that has been explored by many researchers is to represent e-contracts in Event Calculus. Although such representations are intuitive and facilitate temporal reasoning about actions/events and their factual and normative effects, they fall short in situations where domain knowledge cannot be assumed to be complete. Moreover, it is not clear how dynamic normative conflict resolution can be achieved, without resorting to unintuitive representations for conflict resolution strategies. In order to maintain the benefits of an underlying Event Calculus representation, and incorporate assumption-based reasoning and dynamic conflict management capability, we propose a representation of e-contracts as Default Theories, which are constructed by translating Event Calculus representations dynamically. Finally, we discuss how the resulting Default Theory representation enables a software agent to address various reasoning problems.

The usual assumptions of the ant colony meta-heuristic are that each ant constructs its own complete solution and that it will then operate relatively independently of the rest of the colony (with only loose communications via the pheromone structure). However, a more aggressive approach is to allow some measure of competition amongst the ants. Two ways in which this can be done are to allow ants to take components from other ants or limit the number of ants that can make a particular component assignment. Both methods involve a number of competitions so that the probabilistic best assignment of component to ant can be made. Both forms of competitive ant colony optimisation outperform a standard implementation on the benchmark set of the assignment type problem, generalised assignment.

Combinatorial optimization problems are embedded in dynamic environments, spanning many domains. As these problem environments may change repeatedly, agents that attempt to solve problems in such environments must be able to adapt to each change that occurs. We present a technique for a Tabu Search-based meta-heuristic agent that collaborates with a truth maintenance agent to maximize reuse of generated solutions that may become partially inconsistent when a change occurs in the problem space. By allowing the truth maintenance agent to perform partial plan repairs, we hope to mitigate the effect that a change has on the performance of the planning agent. Such a system is discussed in a global logistics scheduling program. The performance of our approach is analyzed with respect to a dynamic constrained vehicle routing problem. Our results show that partial plan repairs increase the stability of solutions in dynamic domains.

Reduced ordered binary decision diagrams (ROBDDs) have become widely used for CAD applications such as logic synthesis, formal verification, and etc. The size of RDBDDs for a Boolean function is very sensitive to the ordering choices of input variables and the problem of finding a minimum-size variable ordering is known to be NP-complete. In this paper, we propose a new ROBDD minimization algorithm based on the microcanonical optimization (MO). MO iteratively applies an initialization phase and a sampling phase to combinatorial optimization problems. In the proposed MO-based algorithm, the initialization phase is replaced with the existing Sifting algorithm known to be a very fast local search algorithm to find a minimum-size ROBDD. We derived equations for the proposed MO-based algorithm parameters empirically. The algorithm has been experimented on well known benchmark circuits and the experiments show that, even with slightly better solutions, the run time of the algorithm is 24% and 48% of the Genetic and SA algorithms’, respectively, on average. The proposed MO-based algorithm is a good candidate for the large size problems that cannot be attacked by exact algorithms when a near-optimal solution is required.

This paper investigates a production planning problem in which the inventory is deteriorating at a constant rate, demand is price dependent and back-order is allowed. It is assumed in most previous works that back-ordering customers may not cancel their orders. However, in reality, many customers may withdraw or cancel their orders before receiving their orders. Tacking the cancellation phenomenon into account, this paper develops a continuous-time model to simultaneously determine the production decision and the selling price. A simple algorithm is used to obtain the optimal solutions. Numerical examples are also used to illustrate the solution searching procedure and the characteristics of the optimal decisions.