Strategy mining is a new area of research about discovering strategies in decision-making. In this paper, we formulate the strategy-mining problem as a clustering problem, called the latent-strategy problem. In a latent-strategy problem, a corpus of data instances is given, each of which is represented by a set of features and a decision label. The inherent dependency of the decision label on the features is governed by a latent strategy. The objective is to find clusters, each of which contains data instances governed by the same strategy. Existing clustering algorithms are inappropriate to cluster dependency because they either assume feature independency (e.g., K-means) or only consider the co-occurrence of features without explicitly modeling the special dependency of the decision label on other features (e.g., Latent Dirichlet Allocation (LDA)). In this paper, we present a baseline unsupervised learning algorithm for dependency clustering. Our model-based clustering algorithm iterates between an assignment step and a minimization step to learn a mixture of decision tree models that represent latent strategies. Similar to the Expectation Maximization algorithm, our algorithm is grounded in the statistical learning theory. Different from other clustering algorithms, our algorithm is irrelevant-feature resistant and its learned clusters (modeled by decision trees) are strongly interpretable and predictive. We systematically evaluate our algorithm using a common law dataset comprised of actual cases. Experimental results show that our algorithm significantly outperforms K-means and LDA on clustering dependency.

In this paper, we describe exploratory experiments for detecting potential "Gray Cygnet" cases that follow a known Black Swan. Gray Cygnets (GCs) are cases that are highly similar and subsequent to novel, surprising, provocative, exceptional cases, so-called Black Swans. They too are surprising, exceptional and provocative in the sense of continuing the change initiated by the Black Swan. Our experiments were carried out using a corpus of common law cases from the United States, particularly New York and Massachusetts, and the United Kingdom primarily in the era 1852-1916 during which there was dramatic change in the prevailing doctrine regarding recovery for damages by a remote buyer. It was provoked by the 1852 landmark case Thomas & Wife v. Winchester and elaborated in subsequent cases. Our methods use similarity measures from Case-Based Reasoning (CBR) based on classic claim lattice and nearest neighbor techniques. We also define and use "supreme on-point cases" (sopc's) that reside in the root node of a claim lattice. We define four simple tests to determine whether a problem case is a potential Gray Cygnet. The tests use the position of a known Black Swan in the analysis of a problem case, specifically whether it appears as a sopc, or mopc, and/or as a nearest neighbor. We also use the tests as an ensemble by using the decision of their majority vote. To evaluate the performance of our tests, we created an answer key based on a closed form definition of a GC in our domain.

The interrupt mechanism in a system-on-chip (SoC) joins the SoCs hardware and software behaviors. We model interrupts as logical rather than physical events and accordingly provides guidelines to compose software components including interrupt-service-routines. As a benefit, classical indeterministic behaviors (due to the parallelism) in the software domain, such as preemption and nesting, can be constructed as early as raw hardware components are being integrated. In effect, while the interrupt mechanism itself is under rigorous stress, it is simultaneously driving the exercise of the entire SoC. This effect can be observed through software profiling at the hardware integration stage.

For IEEE 802.16 broadband wireless access (BWA) system, the support of QoS is of prime importance for providing various traffic profiles in wireless links. Scheduling and resource allocation play the key roles in providing QoS as well as making better use of limited radio resources. This paper proposes an efficient scheduling algorithm for IEEE 802.16 OFDMA system with adaptive allocation of time-frequency resources. We utilize a user and connection based scheme to offer more efficient and QoS-guaranteed performance. The users' signal-to-noise ratio (SNR) on subchannels and their distribution characteristics are also taken into consideration. A new allocation scheme based on maximum deviation channel first (MDCF) is proposed to use the resources more efficiently. The fundamental of allocation scheme is to satisfy the QoS requirements of users as much as possible first, and to maximize system throughput thereafter. The simulation results show that the proposed scheduling and resource allocation algorithm can provide diverse QoS guarantees and use resources efficiently.

The verification of system-on-chip is challenging due to its high level of integration. Multiple components in a system can behave concurrently and compete for resources. Hence, for simulation-based verification, we need a methodology that allows one to automatically generate test cases for testing concurrent and resource-competing behaviors. We introduce the use of a transfer-resource graph (TRG) as the model for test generation. From a high abstraction level, TRG is able to model the parallelism between heterogeneous interaction forms in a system. We show how TRG is used in generating test cases of resource competitions and how these test cases are structured in event-driven test programs. For coverage, TRG can be converted to a Petri net, allowing one to measure the completeness of concurrency in simulation.