Logic Relation Refinement Using Unlabeled Data ∗
نویسندگان
چکیده
The difference in distributions between datasets from different domains, such as different information sources, hinders the direct application of a learned model from one domain to another. We have developed a framework for the adaptation of relational logic models, in particular, Markov Logic Network (MLN), from a source domain to a target domain solving the same task using only unlabeled data in the target domain. In our proposed framework, we modify the model from two aspects, the dependency information across the two domains and within the target domain. First, the relational logic models of the two domains should share certain amount of similarities due to the same goal and similar nature of the data. Hence, we perform model adaptation by penalizing the difference in the two domains and jointly maximizing the likelihood of the target domain and minimizing the difference between the source and the target domain MLNs. Second, closely related information appeared within the target domain is used as additional clues in resolving ambiguous decision making. Potential ambiguity of the model is identified and is refined through incorporating such closely related information. As a result, the adapted model is tailored to the target domain. Our experimental results demonstrate that our adaptation framework is able to improve the performance on the target domain.
منابع مشابه
Algebraic semantics of refinement modal logic
We develop algebraic semantics of refinement modal logic using duality theory. Refinement modal logic has quantifiers that are interpreted using a refinement relation. A refinement relation is like a bisimulation, except that from the three relational requirements only ‘atoms’ and ‘back’ have to be satisfied. We study the dual notion of refinement on algebras and present algebraic semantics of ...
متن کاملIdeal Refinement of Datalog Clauses Using Primary Keys
Inductive Logic Programming (ILP) algorithms are frequently used to data mine multi-relational databases. However, in many ILP algorithms the use of primary key constraints is limited. We show how primary key constraints can be incorporated in a downward refinement operator. This refinement operator is proved to be finite, complete, proper and therefore ideal for clausal languages defined by pr...
متن کاملASM refinement and generalizations of forward simulation in data refinement: a comparison
In [1], we have formalized Börger’s refinement notion for Abstract State Machines (ASMs). The formalization was based on transition systems, and verification conditions were expressed in Dynamic Logic. In this paper, the relation between ASM refinement and data refinement is explored. Data refinement expresses operations and verification conditions using relational calculus. We show how to brid...
متن کاملASMs and Refinement of State-based Systems
My talk will present the refinement concept of abstract state machines (ASMs, [Gur95], [BS03]) that was introduced by Börger in the 90’ies ([Bör90a], [Bör90b]), when he found that traditional refinement concepts were not suitable to capture refinement in the verification of a Prolog compiler. His methodology for the use of the ASM refinement paradigm and numerous applications are described in [...
متن کاملAn analysis of total correctness refinement models for partial relation semantics
This is the second in a series of papers devoted to the thorough investigation of (total correctness) refinement based on an underlying partial relational model. This paper investigates operation refinement and data-refinement based on a weakest precondition interpretation for specifications whose semantics is given by partial relations. We consider three refinement theories based on a weakest ...
متن کامل