Equivalence of models for polynomial learnability

In this paper we consider several variants of Valiant's learnability model that have appeared in the literature. We give conditions under which these models are equivalent in terms of the polynomially learnable concept classes they define. These equivalences allow comparisons of most of the existing theorems in Valiant-style learnability and show that several simplifying assumptions on polynomial learning algorithms can be made without loss of generality. We also give a useful reduction of learning problems to the problem of finding consistent hypotheses, and give comparisons and equivalences between Valiant's model and the prediction learning models of Haussler, Littlestone, and Warmuth (in “29th Annual IEEE Symposium on Foundations of Computer Science,” 1988).     

Difficult to Let Housing Association Stock in England: Property,Management and Context

Problems related to unpopular housing are not new in Britain's social rented sector. In the past, however, these have generally been associated with housing provided directly by the state through local authorities. For many, the term 'difficult to let' has become synonymous with large, ageing, poor quality estates managed under a centralised framework and subject to a long period of low investment and indifferent management. There has also been a long-standing assumption that social housing allocations systems tend to channel the most vulnerable tenants into the least desirable housing. In fact, the problem of difficult to let property is now as pervasive in the not-for-profit sector managed by housing associations as it is within council housing. In many cases the developments affected are small in scale and, in a significant proportion of cases, recently constructed. In explaining why property becomes difficult to let, the local and regional context is clearly important. The balance between social, economic and housing demand circumstances and property and estate characteristics varies between regions. There is relatively little evidence that difficult to let housing association properties are occupied by disproportionate numbers of the most severely disadvantaged tenants and this suggests that something has happened to modify the segregating effects of allocations systems observed in earlier studies. Nevertheless, this apparent change must be seen within the context of the general trend towards a residual role for social housing in its entirety. The social class and income range of new social sector tenants now is much diminished by comparison with earlier decades with few distinctions between the occupiers of difficult to let and other housing.     

Polyurethane‐infiltrated carbon foams: A coupling of thermal and mechanical properties

The thermal and mechanical properties of polyurethane‐infiltrated carbon foam of various densities were investigated. By combining the high thermal conductivity of the carbon foam with the mechanical toughness of the pure polyurethane, a mechanically tough composite (relative to the unfilled foam) that could be used at higher temperatures than the polyurethane's degradation was formed. Both the tensile strength and the modulus increased by an order of magnitude for the composites compared to unfilled foam, while the compressive and shear strengths and moduli of the composites approached values exhibited by pure polyurethane. At both 300 and 400°C, the rectangular blocks of pure polyurethane lost their mechanical integrity due to decomposition in air. Thermogravimetric analysis confirms substantial initial weight loss above 290°C. Filled carbon foam blocks, however, maintain their mechanical integrity at both 300 and 400°C indefinitely, although the bulk of the rectangular block mass is polyurethane. Three different carbon foam densities are examined. As expected, the higher density foams show greater heat transfer. © 2003 Wiley Periodicals, Inc. J Appl Polym Sci 87: 2348–2355, 2003     

An Experimental and Theoretical Comparison of Model Selection Methods

Machine Learning - We investigate the problem of model selection in the setting of supervised learning of boolean functions from independent random examples. More precisely, we compare methods for...     

Neighbourhood Restructuring and Residential Relocation: Towards a Balanced Perspective on Relocation Processes and Outcomes

This introductory paper to this special issue of Housing Studies questions whether various characteristics of the debate and research on gentrification, displacement and restructuring justify a largely negative perspective on the processes and outcomes of ‘forced’ residential relocation. We argue that a proper and fuller consideration of issues around policy, context, process and outcomes enable researchers and commentators to avoid ready characterisations and self-fulfilling investigations of restructuring which serve to present it as a singular (and somewhat suspicious or conspiratorial) phenomenon. For this purpose, we present a broad conceptual framework for restructuring and relocation studies, based on these four themes. Subsequently, we review major issues in restructuring and gentrification discourses, and briefly reflect upon some of the factors underlying the negative loading of the term displacement. We also identify caveats in the evidence base of relocation studies, both in the United States and in Europe. Finally, we introduce the papers in this special issue. The overall aim of this issue is to offer a more open, balanced starting position for analysis of urban restructuring processes and relocation outcomes, particularly in relation to areas of social housing.     

The place of neighbourhood in social cohesion: insights from massey,West Auckland

Within public health discourse there is growing recognition that the social structures and ecology of neighbourhoods impact on health and wellbeing. A clearer understanding of the pathways through which this influence occurs will inform locality-based interventions to tackle inequalities. In this paper we report on an interview based study with parents of young children living in Massey, a suburban neighbourhood of metropolitan Auckland. Locational access to community resources is comparatively poor in Massey as measured by a community resource accessibility index. The findings provide insights into the impact of such access to services and amenities on neighbourhood social cohesion. We conclude by suggesting planning strategies for strengthening neighbourhood identification and attachment for this population group.     

A Sparse Sampling Algorithm for Near-Optimal Planning in Large Markov Decision Processes

A critical issue for the application of Markov decision processes (MDPs) to realistic problems is how the complexity of planning scales with the size of the MDP. In stochastic environments with very large or infinite state spaces, traditional planning and reinforcement learning algorithms may be inapplicable, since their running time typically grows linearly with the state space size in the worst case. In this paper we present a new algorithm that, given only a generative model (a natural and common type of simulator) for an arbitrary MDP, performs on-line, near-optimal planning with a per-state running time that has no dependence on the number of states. The running time is exponential in the horizon time (which depends only on the discount factor and the desired degree of approximation to the optimal policy). Our algorithm thus provides a different complexity trade-off than classical algorithms such as value iteration—rather than scaling linearly in both horizon time and state space size, our running time trades an exponential dependence on the former in exchange for no dependence on the latter.Our algorithm is based on the idea of sparse sampling. We prove that a randomly sampled look-ahead tree that covers only a vanishing fraction of the full look-ahead tree nevertheless suffices to compute near-optimal actions from any state of an MDP. Practical implementations of the algorithm are discussed, and we draw ties to our related recent results on finding a near-best strategy from a given class of strategies in very large partially observable MDPs (Kearns, Mansour, & Ng. Neural information processing systems 13, to appear).     

Near-Optimal Reinforcement Learning in Polynomial Time

We present new algorithms for reinforcement learning and prove that they have polynomial bounds on the resources required to achieve near-optimal return in general Markov decision processes. After observing that the number of actions required to approach the optimal return is lower bounded by the mixing time T of the optimal policy (in the undiscounted case) or by the horizon time T (in the discounted case), we then give algorithms requiring a number of actions and total computation time that are only polynomial in T and the number of states and actions, for both the undiscounted and discounted cases. An interesting aspect of our algorithms is their explicit handling of the Exploration-Exploitation trade-off.     

Infection of human marrow stroma by human immunodeficiency virus-1 (HIV-1) is both required and sufficient for HIV-1-induced hematopoietic suppression in vitro: demonstration by gene modification of primary human stroma

Patients with human immunodeficiency virus-1 (HIV-1) infection often present with bone marrow (BM) failure that may affect all hematopoietic lineages. It is presently unclear whether this failure reflects a direct viral impairment of the CD34+ hematopoietic progenitor cells or whether the virus affects the BM microenvironment. To study the effects of HIV-1 on the BM microenvironment, we examined the stromal cell monolayers in long-term BM culture (LTBMC), which are the in vitro equivalent of the hematopoietic microenvironment. We assessed the hematopoietic support function (HSF) of human stromal layers by determining the cellular proliferation and colony-forming ability of hematopoietic progenitors from BM cells grown on the stromal layers. We show that the HSF is reduced by in vitro infection of the human stromal cell layer by a monocytotropic isolate of HIV-1 (JR-FL). There is no loss of HSF when the stromal cell layer is resistant to HIV-1 replication, either using murine stromal cell layers that are innately resistant to HIV-1 infection or using human stromal cells genetically modified to express a gene that inhibits HIV-1 replication (an RRE decoy). Decreased HSF was seen using either human or murine hematopoietic cells, if the stromal cells were human cells that were susceptible to HIV-1 infection. These in vitro studies implicate HIV-1 replication in the stroma as the essential component causing decreased hematopoietic cell production in HIV-1 infection.