Learning from examples with unspecified attribute values

A challenging problem within machine learning is how to make good inferences from data sets in which pieces of information are missing. While it is valuable to have algorithms that perform well for specific domains, to gain a fundamental understanding of the problem, one needs a “theory” about how to learn with incomplete data. The important contribution of such a theory is not so much the specific algorithmic results, but rather that it provides good ways of thinking about the problem formally. In this paper we introduce the unspecified attribute value (UAV) learning model as a first step towards a theoretical framework for studying the problem of learning from incomplete data in the exact learning framework.In the UAV learning model, an example x is classified positive (resp., negative) if all possible assignments for the unspecified attributes result in a positive (resp., negative) classification. Otherwise the classification given to x is “?” (for unknown). Given an example x in which some attributes are unspecified, the oracle UAV-MQ responds with the classification of x. Given a hypothesis h, the oracle UAV-EQ returns an example x (that could have unspecified attributes) for which h(x) is incorrect.We show that any class of functions learnable in Angluin’s exact model using the MQ and EQ oracles is also learnable in the UAV model using the MQ and UAV-EQ oracles as long as the counterexamples provided by the UAV-EQ oracle have a logarithmic number of unspecified attributes. We also show that any class learnable in the exact model using the MQ and EQ oracles is also learnable in the UAV model using the UAV-MQ and UAV-EQ oracles as well as an oracle to evaluate a given boolean formula on an example with unspecified attributes. (For some hypothesis classes such as decision trees and unate formulas the evaluation can be done in polynomial time without an oracle.) We also study the learnability of a universal class of decision trees under the UAV model and of DNF formulas under a representation-dependent variation of the UAV model.     

Dual contextual module for neural machine translation

Machine Translation - Self-attention-based encoder-decoder frameworks have drawn increasing attention in recent years. The self-attention mechanism generates contextual representations by attending...     

Clustering semantically heterogeneous distributed aggregate databases

Databases developed independently in a common open distributed environment may be heterogeneous with respect to both data schema and the embedded semantics. Managing schema and semantic heterogeneities brings considerable challenges to learning from distributed data and to support applications involving cooperation between different organisations. In this paper, we are concerned mainly with heterogeneous databases that hold aggregates on a set of attributes, which are often the result of materialised views of native large-scale distributed databases. A model-based clustering algorithm is proposed to construct a mixture model where each component corresponds to a cluster which is used to capture the contextual heterogeneity among databases from different populations. Schema heterogeneity, which can be recast as incomplete information, is handled within the clustering process using Expectation-Maximisation estimation and integration is carried out within a clustering iteration. Our proposed algorithm resolves the schema heterogeneity as part of the clustering process, thus avoiding transformation of the data into a unified schema. Results of algorithm evaluation on classification, scalability and reliability, using both real and synthetic data, demonstrate that our algorithm can achieve good performance by incorporating all of the information from available heterogeneous data. Our clustering approach has great potential for scalable knowledge discovery from semantically heterogeneous databases and for applications in an open distributed environment, such as the Semantic Web.     

Adapting Memory Hierarchies for Emerging Datacenter Interconnects

E?cient resource utilization requires that emerging datacenter interconnects support both high performance communication and e?cient remote resource sharing. These goals require that the network be mor...     

Midlatitude cirrus cloud climatology from the Facility for Atmospheric Remote Sensing. IV. Optical displays

In this fourth of a series of papers that describe long-term cloud research at the Facility for Atmospheric Remote Sensing at Salt Lake City, Utah, an approximately 10-year record of polarization lidar and photographic observations is analyzed to characterize the occurrence of optical displays in our local varieties of midlatitude cirrus clouds. The frequencies of occurrence of various types of halo, arc, and corona displays are evaluated according to their appearance and longevity over nominal 1-h observation periods and to the meteorological source of the cirrus. We find that complex halo-arc displays are rare at our locale and that even the so-called common 22 degree halo occurs infrequently as a complete long-lived ring. For example, only approximately 6% of the 1561-h daytime cirrus periods have bright and prolonged 22 degree halos, although a total of 37.3% have some indications of this halo, even if they are brief and fragmentary. Other fairly frequent features are the 22 degree upper tangent arc (8.6%), 22 degree parhelia (8.5%), and solar corona (7.2%). Of the optical displays observed, 83.6% are refraction based, only 1.9% are due to reflection phenomena, and a surprising 15.4% are caused by diffraction. Complex halo-arc displays are disproportionally associated with cirrus formed in tropical or subtropical airflow and also contain more horizontally oriented planar ice crystals. Lidar linear depolarization ratios from a subset of vivid displays show significant differences between halo- and the corona-producing cirrus, reflecting the effects of particle shape. Halos are associated with relatively warm cirrus that contain randomly and horizontally oriented planar ice crystals, whereas the colder corona cirrus produce much stronger depolarization from crystals too small to be uniformly oriented. Comparisons are made with available information from other locales, and we attempt to explain the geographical differences in terms of basic cirrus cloud processes.     

Using Theory to Inform Capacity‐Building: Bootstrapping Communities of Practice in Computer Science Education Research

In this paper, we describe our efforts in the deliberate creation of a community of practice of researchers in computer science education (CSEd). We understand community of practice in the sense in which Wenger describes it, whereby the community is characterized by mutual engagement in a joint enterprise that gives rise to a shared repertoire of knowledge, artefacts, and practices. We first identify CSEd as a research field in which no shared paradigm exists, and then we describe the Bootstrapping project, its metaphor, structure, rationale, and delivery, as designed to create a community of practice of CSEd researchers. Features of other projects are also outlined that have similar aims of capacity building in disciplinary‐specific pedagogic enquiry. A theoretically derived framework for evaluating the success of endeavours of this type is then presented, and we report the results from an empirical study. We conclude with four open questions for our project and others like it: Where is the locus of a community of practice? Who are the core members? Do capacity‐building models transfer to other disciplines? Can our theoretically motivated measures of success apply to other projects of the same nature?     

Field evidence of biodegradation of N-Nitrosodimethylamine (NDMA) in groundwater with incidental and active recycled water recharge

Biodegradation of N-Nitrosodimethylamine (NDMA) has been found through laboratory incubation in unsaturated and saturated soil samples under both aerobic and anaerobic conditions. However, direct field evidence of in situ biodegradation in groundwater is very limited. This research aimed to evaluate biodegradation of NDMA in a large-scale groundwater system receiving recycled water as incidental and active recharge. NDMA concentrations in 32 monitoring and production wells with different screen intervals were monitored over a period of seven years. Groundwater monitoring was used to characterize changes in the magnitude and extent of NDMA in groundwater in response to seasonal hydrogeologic conditions and, more importantly, to significant concentration variations in effluent from water reclamation plants (associated with treatment-process changes). Extensive monitoring of NDMA concentrations and flow rates at effluent discharge locations and surface-water stations was also conducted to reasonably estimate mass loading through unlined river reaches to underlying groundwater. Monitoring results indicate that significant biodegradation of NDMA occurred in groundwater, accounting for an estimated 90% mass reduction over the seven-year monitoring period. In addition, a discrete effluent-discharge and groundwater-extraction event was extensively monitored in a well-characterized, localized groundwater subsystem for 626 days. Analysis of the associated NDMA fate and transport in the subsystem indicated that an estimated 80% of the recharged mass was biodegraded. The observed field evidence of NDMA biodegradation is supported by groundwater transport modeling accounting for various dilution mechanisms and first-order decay for biodegradation, and by a previous laboratory study on soil samples collected from the study site [Bradley, P.M., Carr, S.A., Baird, R.B., Chapelle, F.H., 2005. Biodegradation of N-Nitrosodimethylamine in soil from a water reclamation facility. Bioremediat. J. 9 (2), 115-120.].     

Development of children's understanding of ambivalence and causal theories of emotions.

Investigated a 4-stage developmental sequence in children's understanding of ambivalence. At the highest level of this sequence, children understand that contradictory feelings can coexist at the same time toward the same target in terms of a framework in which it is recognized that the 2 feelings influence one another. 20 4–5 yr olds, 20 7–8 yr olds, and 20 10–11 yr olds were presented with 2 tape-recorded stories in which the protagonist could be construed as feeling happy and sad or anger and love. Raters scored the Ss' responses to a structured interview about the stories in terms of the proposed sequence. Separate parts of the protocols were assessed independently with respect to a proposed sequence in children's causal theories of emotions. According to this sequence, children's understanding of feelings develops from an externally based theory in which feelings are wedded to events to an internally based theory in which feelings largely depend on memories, thoughts, and attitudes. Results support both hypothesized developmental progressions and show significant relationships between the 2 domains of Ss' understanding of emotions. Findings suggest that internal states' mediation of emotional responses may be an important component in understanding ambivalence and that the motivation to integrate conflicting feelings may play a role in promoting the external to internal shift in children's causal theories of emotions. (29 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

The claiming effect: Why players are more generous in social dilemmas than in ultimatum games.

The term procedural frames is introduced and defined as different representations of structurally equivalent allocation processes. Study 1 compared 2 well-known games, sequential social dilemmas and ultimatum bargaining, that share the same structure: Player 1 creates an allocation of a resource and Player 2 decides whether to allow it or deny it. Study 1 found that Player 1 made more favorable allocations and Player 2 accepted more unfavorable allocations in a social dilemma frame than in an equivalent ultimatum bargaining frame. Study 2 revealed the critical determinant was whether Player 2 had to respond to an allocation by accepting or rejecting it (as in the ultimatum game) or by making a claim (as in the social dilemma). Two additional studies explored how these actions are perceived. The inconsistency of behavior across procedural frames raises methodological concerns but illuminates construal processes that guide allocation. (PsycINFO Database Record (c) 2010 APA, all rights reserved)