Fu-Ming Lee

A study on dynamic bargaining strategy under time constraints and with incomplete information

On the Internet, the bilateral bargaining agents would often fall into failure or cause poor utility value at the end of bargaining under time constraints and with incomplete information. The core issue of bargaining between agents on the Internet is the bargaining strategies. However, under time constraints and with incomplete information, the bilateral agents tend to obtain poor utility value. This outcome usually causes the fail bargaining. To overcome the problem, this paper proposes a dynamic bargaining algorithm with novel and simple characteristics to make offers. The algorithm predicts the opponents possible offers for the consecutive rounds with little information about its opponents deadline, reservation price, and bargaining strategy. Then, it revises its offer function to maximize its utility in one possible agreement point. The experiments show that the algorithm would enhance the ratio of reaching agreement and the utility value at the end of the bargaining.

Using Ontology for Personalized Mobile Message Computation

With the convenient of mobile device, mobile users are able to “pull” or “push” the mobile message at anytime and anywhere. The power of mobile message has turned into an important marketing power. However, how to utilize the availability and the quick-viewing convenience of mobile message to match the personal needs are not discussed. From the past studies, we notice that the studies of mobile recommendation focus more on location service and push service; few studies are focusing on personal mobile-message recommendation and its performance. In order to provide the personalized services, this research proposes three computation processes for conducting the mobile message recommendation. We apply ontology to construct the user preference profile and the message template containing the product information or service information. For personalized recommendation, the user preference profile and the message template are compared. The ontology matching techniques are proposed, they are: (1) Breadth-First-Matching (BFM), (2) Depth-First-Matching (DFM), and (3) Node-Index-Matching (NIM). The experiments are designed for examining the message “pull” service. To prove the proposed matching computations are applicable, the evaluation metrics use matching count for performance comparison. The experimental results show that the Node-Index-Matching (NIM) outperforms the other two matching methods and is good for the mobile message recommendation.

Stable on-line parameter identification algorithms for systems with non-parametric uncertainties and disturbances

An on-line parameter identification problem is formulated for linear time-invariant continuous-time systems with bounded input/output disturbances as well as non-parametric uncertainties characterized either by H2 or H∞ norms. Based on the formulation, a switching type gradient algorithm is proposed to estimate the parameters of the system from the available input-output data. In spite of the existence of non-parametric uncertainties and disturbances, this on-line algorithm guarantees that the estimation error is monotonically decreasing with respect to time, and the parameter estimate is convergent to a steady-state value under a mild condition. Furthermore, the algorithm is stable in the sense that the estimation error will converge to zero as both non-parametric uncertainties and disturbances gradually diminish. To evaluate the accuracy of the identified parameters, an upper bound on the estimation error is given.

Robust on-line parameter identification with general knowledge on level of information noise: continuous and discrete cases

A robust on-line parameter identification problem is posed and solved for systems with general knowledge of the level of the inherent information noise. Both continuous-time and discrete-time cases are considered in this paper. For the former case, the knowledge can be the bound on either the magnitude or the finite-time L/sup p/ norm, p/spl isin/[1, /spl infin/), of the noise. Whereas for the latter case, it can be the bound on either the magnitude or the finite-index l/sup p/ norm, p/spl isin/[1, /spl infin/), of the noise. Based on the knowledge, a switching type algorithm is proposed to estimate the parameters of the system from the available input-output data. In spite of the existence of the information noise, this on-line algorithm guarantees that the estimation error is monotonically decreasing, and the parameter estimate is convergent to a steady state value under a mild condition.

Set-membership identification for continuous-time systems with nonparametric uncertainties and disturbances

An online set-membership identification problem is formally formulated for linear time-invariant SISO continuous-time systems which have bounded disturbances as well as nonparametric uncertainties. Based on the formulation, an efficient ellipsoidal-bounding algorithm is proposed to estimate the parameter set of the system from the available input-output timed data.<<ETX>>

A Personalized Pair-Recommendation Approach Using Mobile Message Ontology

Nowadays, many commercial products or services are promoted through the mobile device by using the mobile message (MS). How to provide proper and suitable MS to cope with user¡¦s preference is an important issue for business. To understand and infer the user message usage or user viewing behavior, ontology can be applied to conceptualize the user¡¦s preference and to construct the personal message profile. By incorporating the recommendation method, the mobile message can be recommended in terms of personalization. To increase the MS viewing ratio, this research proposed a Personalized Pair-Recommendation (PPR) method to provide not only one but related message for satisfying user¡¦s need. Our proposed PPR will analyze the mobile user¡¦s preference using the ontology of user¡¦s message preference. To illustrate the usefulness of our proposed method, we tested both Content-based (CB) method and Collaborative Filtering (CF) method, two major types of recommendation, to examine the precision, F1-measure, and the Successful Rate (SR). According to the results of the experiments, the proposed Personalized Pair-Recommendation (PPR) when integral with CF method can produce better outcome in terms of SR and F1-Measure.

Stable on-line parameter identification with general knowledge on level of information noise: discrete-time case

Abstract An on-line parameter identification problem is posed and solved for discrete-time systems with general knowledge on the level of the inherent information noise. The knowledge can be the bound on either the magnitude or the finite-index l p norm, pϵ [1, ∞), of the noise. Based on the knowledge, a switching type gradient algorithm (or called gradient algorithm with dead zone) is proposed to estimate the parameters of the system from the available input-output data. In spite of the existence of the noise, this on-line algorithm guarantees that the estimation error is monotonically decreasing, and the parameter estimate is convergent to a steady-state value under a mild condition. Furthermore, the algorithm is stable in the sense that the estimation error will converge to zero as the bound on the noise gradually diminishes.

Convergent On-Line Identification in H ∞ for Discrete Systems

Abstract Two related on-line identification problems are formulated and solved for stable discrete systems. The first problem involves identification of frequency responses of a system at an arbitrary finite set of frequency points from the available input-output timed data. The second problem involves identification of the system model in H ∞ also from the available input-output timed data. Concrete convergent algorithms are specified for these two problems. Each of these algorithms provides on-line not only an estimate of the system but also an explicit upper bound on the worst-case deterministic identification error.