Fact | EPFL Graph Search

A fact is a true datum about one or more aspects of a circumstance. Standard reference works are often used to check facts. Scientific facts are verified by repeatable careful observation or measurement by experiments or other means. For example, "This sentence contains words." accurately describes a linguistic fact, and "The sun is a star" accurately describes an astronomical fact. Further, "Abraham Lincoln was the 16th President of the United States" and "Abraham Lincoln was assassinated" both accurately describe historical facts. Generally speaking, facts are independent of belief and of knowledge and opinion. Facts are different from inferences, theories, values, and objects. Etymology and usage The word fact derives from the Latin factum. It was first used in English with the same meaning: "a thing done or performed" a meaning now obsolete. The common usage of "something that has really occurred or is the case" dates from the mid-16th century.

Optimalisation des formes d'organisation dans l'industrie de la construction

Philippe Pot

A piece of work carried out by the construction industry, whether a building or a civil engineering project, is not a mass-produced item. It is always a unique piece of work, by dint of its response to the specific needs of the owner and its potential users, its integration or otherwise into a built environment, and its construction on ground whose geo-technical properties are always specific. This unique piece of work, carried out outside on a site subject to numerous hazards, must nevertheless meet the programme's fixed objectives first time round. Unfortunately, unlike most industrial products, it is usually, even nowadays, the outcome of a fragmented approach: designers and builders of the shell of the building and those engaged in the finishings are in fact involved in a segmented and sequential process, without any real interaction between the various disciplines, to the detriment of overall quality and the cost of the end-product. The author of this thesis has had the opportunity, in his professional career in the management of major construction companies, of testing most organisational forms in the sector. He has observed that, depending on the model chosen, they exercise a more or less beneficial effect on quality, costs and time-frames for completing the works, as well as on the capacity for innovation of the participants themselves. The first objective of the thesis was to analyse the strengths and weaknesses of the most common organisational models in the construction industry and to identify factors in the success of the construction process. The second objective was to suggest an original organisational model which would make optimum use of these factors, taking into account modern organisational theory and project management. The first section of the thesis is an historical study of organisational forms of construction from antiquity to the modern times, and examines in particular the relationships between those actively and directly involved (owner, designers and builders) and their relationships in turn with those involved indirectly (users, public administration, finance providers, environmental and heritage protection organisations, laws and standards, etc.). This historical perspective makes a useful contribution because, although much has been written on architecture and architects, especially the most famous among them, and sometimes on engineers, work looking systematically at the organisational forms used in the construction industry over the centuries is rare. This study demonstrates that the fragmented organisation of construction characteristic of the 20th century is merely an accident in the long history of construction; the twenty-five preceding centuries almost always favoured close collaboration between designers and builders and the wonderful works they left behind demonstrate the wisdom of this approach. Study of the history also provides us with a list of the main success factors and the demands to which the successful organisation of construction must respond. The second part endeavours to verify the relevance of the needs drawn, in the light of modern organisational theory, from project management and creativity, and to complete the list. It goes on to analyse on that basis the strengths and weaknesses of the organisational models most commonly used in the modern construction industry and demonstrates that organisation as a design and build, is in the best position at this stage in the research to meet the needs of owners. The third section verifies, through case studies of five pieces of work recently completed or under construction, whether this organisational model effectively satisfies the expectations of those involved in the construction project, simultaneously examining potential improvements. To that end, a series of interviews was conducted in each case study, with a view to eliciting the views of representative participants: owner, operator, user, project manager, architect, engineer and sub-contracting trades. The process was completed with an analysis of the attitude of some public sector owners to the design and build organisational model. Finally, the lessons drawn from these case studies were used to complete the list of success factors already identified. A final section sets out an original organisational model named Integrated Construction Management (ICM), which recommends an inter-disciplinary approach to construction and responds to the success factors identified in the first three sections. It involves a new division of the production process, which provides the owner with a broader and more cost-effective range of innovative architectural and construction options. It also delivers a series of practical recommendations to assist in the application of the proposed model. In conclusion, Integrated Construction Management places the owner at the centre of a transparent and progressive production process, simultaneously removing the majority of the risks involved in construction. Through its inter-disciplinary approach, it offers innovative solutions together with improvements to the overall quality of the piece of work itself on favourable economic terms. Finally, it enables the owner to take informed decisions and to use his or her financial resources as and when the feasibility of the project is confirmed. It is hoped that, as a result, these advantages will be demonstrated through pilot projects carried out using the ICM model, which will without doubt make possible further improvements in performance.

EPFL2005

Theory of representation learning in cortical neural networks

Carlos Stein Naves de Brito

Our brain continuously self-organizes to construct and maintain an internal representation of the world based on the information arriving through sensory stimuli. Remarkably, cortical areas related to different sensory modalities appear to share the same functional unit, the neuron, and develop through the same learning mechanism, synaptic plasticity. It motivates the conjecture of a unifying theory to explain cortical representational learning across sensory modalities. In this thesis we present theories and computational models of learning and optimization in neural networks, postulating functional properties of synaptic plasticity that support the apparent universal learning capacity of cortical networks. In the past decades, a variety of theories and models have been proposed to describe receptive field formation in sensory areas. They include normative models such as sparse coding, and bottom-up models such as spike-timing dependent plasticity. We bring together candidate explanations by demonstrating that in fact a single principle is sufficient to explain receptive field development. First, we show that many representative models of sensory development are in fact implementing variations of a common principle: nonlinear Hebbian learning. Second, we reveal that nonlinear Hebbian learning is sufficient for receptive field formation through sensory inputs. A surprising result is that our findings are independent of specific details, and allow for robust predictions of the learned receptive fields. Thus nonlinear Hebbian learning and natural statistics can account for many aspects of receptive field formation across models and sensory modalities. The Hebbian learning theory substantiates that synaptic plasticity can be interpreted as an optimization procedure, implementing stochastic gradient descent. In stochastic gradient descent inputs arrive sequentially, as in sensory streams. However, individual data samples have very little information about the correct learning signal, and it becomes a fundamental problem to know how many samples are required for reliable synaptic changes. Through estimation theory, we develop a novel adaptive learning rate model, that adapts the magnitude of synaptic changes based on the statistics of the learning signal, enabling an optimal use of data samples. Our model has a simple implementation and demonstrates improved learning speed, making this a promising candidate for large artificial neural network applications. The model also makes predictions on how cortical plasticity may modulate synaptic plasticity for optimal learning. The optimal sampling size for reliable learning allows us to estimate optimal learning times for a given model. We apply this theory to derive analytical bounds on times for the optimization of synaptic connections. First, we show this optimization problem to have exponentially many saddle-nodes, which lead to small gradients and slow learning. Second, we show that the number of input synapses to a neuron modulates the magnitude of the initial gradient, determining the duration of learning. Our final result reveals that the learning duration increases supra-linearly with the number of synapses, suggesting an effective limit on synaptic connections and receptive field sizes in developing neural networks.

EPFL2016

Turnover, account value and diversification of real traders: evidence of collective portfolio optimizing behavior

Damien Challet, David Sébastien Morton de Lachapelle

Despite the availability of very detailed data on financial markets, agent-based modeling is hindered by the lack of information about real trader behavior. This makes it impossible to validate agent-based models, which are thus reverse-engineering attempts. This work is a contribution towards building a set of stylized facts about the traders themselves. Using the client database of Swissquote Bank SA, the largest online Swiss broker, we find empirical relationships between turnover, account values and the number of assets in which a trader is invested. A theory based on simple mean-variance portfolio optimization that crucially includes variable transaction costs is able to reproduce faithfully the observed behaviors. We finally argue that our results bring to light the collective ability of a population to construct a mean-variance portfolio that takes into account the structure of transaction costs.