#### DMCA

## Acquiring Symbolic Design Optimization Problem Reformulation Knowledge (2009)

### Citations

2848 |
Pattern Classification.
- Duda, Hart, et al.
- 2001
(Show Context)
Citation Context ...asurement and clustering techniques in all these domains. Whilesdimensionality reduction, similarity measurements and clustering are commonly adoptedsapproaches in machine learning and AI algorithms (=-=Duda et al., 2000-=-; Russell & Norvig,s2003), it is the role of SVD that is mathematically intriguing and interesting. Despite beingsused in such diverse domains for pattern extraction, it is not a commonly discussed me... |

2521 |
The Sciences of the Artificial
- Simon
- 1996
(Show Context)
Citation Context ... knowledge of the world by employing these abstractions throughsrepresentations.sDesign is a discipline with the pragmatic aim of transforming the world we inhabit into asmore desirable future world (=-=Simon, 1969-=-/ 1981). Design deals with a vision that has to besdescribed and represented in order to be realized. This act of describing is a challenge becauseswhat is being described does not exist yet. Design m... |

1936 | Information theory, inference, and learning algorithms
- MacKay
- 2003
(Show Context)
Citation Context ...sed. The K-means clustering algorithm issan iterative method for putting N data points in an I-dimensional space into K clusters, whereseach cluster is parameterized by a vector m(k) called its mean (=-=Mackay, 2003-=-). The dataspoints are vectors denoted by x(n), where n = 1:N, each x has I components xi. Each dataspoint is assigned to the nearest mean (or centroid). A metric is defined for measuringsdistances be... |

1816 | A solution to Plato's problem: The latent semantic analysis theory of acquisition, induction and representation of knowledge
- Landauer, Dumais
- 1997
(Show Context)
Citation Context ...sition (SVD) in various knowledge domains to address the specificsproblem of extracting semantic patterns from a syntactic representation. In SNLP, the LatentsSemantic Analysis (LSA) method uses SVD (=-=Landauer & Dumais, 1997-=-) to reveal semanticspatterns in textual data. The underlying idea in LSA is that linguistic knowledge contains aslarge number of weak correlations between semantic concepts, and that these latentscor... |

1225 | Probabilistic latent semantic indexing,”
- Hofmann
- 1999
(Show Context)
Citation Context ... method that provides a range of “good” solutions. A seconds134 approach to consider (in addition to the one presented in Section 10.3.1.1) would be thesProbabilistic Latent Semantic Approach (PLSA) (=-=Hofmann, 1999-=-) that approaches the samesproblem from a statistical foundation. In the PLSA approach, occurrence data from anysdyadic domain (but most noticeably from the language domain, i.e. words and documents) ... |

676 | Using Linear Algebra for Intelligent Information Retrieval,
- Berry
- 1995
(Show Context)
Citation Context ... are vectors that can be assessed for similarity in terms of some distance metric.sUsually, similarity is assessed between them through cosine distance measurements. Figures2.7 shows an example from (=-=Berry & Dumais, 1994-=-) showing how words and documents aresrepresented in a k=2 space and a query in this space that retrieves the words and documentsswithin a certain cosine threshold. A higher cosine means higher simila... |

369 |
The symbolic species: The co-evolution of language and the brain.
- Deacon
- 1997
(Show Context)
Citation Context ...able to “see”sthis.s3.3.2s“Symbols aren’t simple”: Why a purely syntactical analysis reveals “meaning”sA symbol is, in a way, the result of the “highest level” of abstraction produced by any speciess(=-=Deacon, 1997-=-) over perceptual conceptual dynamic activations in experiences. “Highest”sdoes not imply a gradation of intelligence, only that the semiotic mode of reference andscommunication enables humans to demo... |

364 |
Situated Cognition: On Human Knowledge and Computer Representations.
- Clancey
- 1997
(Show Context)
Citation Context ...ind) is symbolics(or sub-symbolic as claimed by the neural networks/ connectionist approach).sWhether knowledge in the mind exists in symbolic form (Simon, 1995), or is onlysrepresented symbolically (=-=Clancey, 1997-=-, 1999) is a lasting debate relevant to any disciplinesthat concerns itself with intelligence and its artificial construction. Design is no exception. Insdesign, this debate becomes particularly inter... |

353 |
Introduction to Linear Algebra,
- Strang
- 1998
(Show Context)
Citation Context ...SVD and the correct choice of dimensionality to view these relations aresable to reveal the semantic patterns and the “latent” semantic meaning of the text. In digitalsimage processing (Kalman, 1996; =-=Strang, 2003-=-), SVD is used as a mathematical tool tosidentify pattern redundancy in image data for compression of images. Further, this researchsfound that SVD, as an algorithm, has the capacity to be used as a g... |

227 |
From sensation to cognition.
- Mesulam
- 1998
(Show Context)
Citation Context ... thesspatial and temporal context in which the experience occurred, and semantic memory issdefined as the use of general and invariant facts about the world as abstractions over episodicsexperiences (=-=Mesulam, 1998-=-). In empirical cognitive neuroscience, observation experimentssusing functional magnetic resonance imaging (fMRI) on human subjects report that the samespart of the brain is shown as active when subj... |

165 |
Computer Science as Empirical Enquiry: Symbols and Search.
- Newell, Simon
- 1976
(Show Context)
Citation Context ...ld be beneficial tosdevelop computational methods that could assist the designer with problem modeling andsreformulation tasks.s5 1.1.2 Design theory motivationsThe physical symbol system hypothesis (=-=Newell & Simon, 1976-=-; Simon, 1995) establishes asprevalent methodological and theoretical approach for developing AI algorithms. This is thesexplicit symbolic reasoning approach (Cagan et al., 1997), which specifies that... |

145 | Integration Analysis of Product Decompositions, In:
- Pimmler, Eppinger
- 1994
(Show Context)
Citation Context ...em (Michelena & Papalambros, 1997), an optimal design problem (Papalambross& Wilde, 2000), a standard non-analytic Functional Dependence Table (FDT) or a DesignsStructure Matrix (DSM) (Li & Li, 2005; =-=Pimmler & Eppinger, 1994-=-). A symbol in a designsmodel represents a structure or behavior variable (for e.g. internal diameter, force), parameters(e.g. minimum wall thickness), or system component (e.g. Fan Blade in an aeroen... |

125 | Unexpected discoveries and S-invention of design requirements: important vehicles for a design process.
- Suwa, Gero, et al.
- 2000
(Show Context)
Citation Context ...or linguistic modes of designing, the representationsitself is ambiguous and the encoding of ambiguity is considered a robust mechanism forsaccommodating multiple active ideas (Schon & Wiggins, 1992; =-=Suwa et al., 2000-=-), thesmathematical modeling of designed objects does not lend itself to the same kind of ambiguity.sIn this mode of designing, the construction of a representation is a precise, well-structured,sunam... |

119 |
Kinds of Seeing and their Functions
- Schön, Wiggins
- 1992
(Show Context)
Citation Context ...s in visual, graphical or linguistic modes of designing, the representationsitself is ambiguous and the encoding of ambiguity is considered a robust mechanism forsaccommodating multiple active ideas (=-=Schon & Wiggins, 1992-=-; Suwa et al., 2000), thesmathematical modeling of designed objects does not lend itself to the same kind of ambiguity.sIn this mode of designing, the construction of a representation is a precise, we... |

99 |
Principles of Optimal Design,
- Papalambros, Wilde
- 2000
(Show Context)
Citation Context ... automate because the task issdependent on interaction with the design problem, knowledge intensive, requires humanssubjective decision-making and a large amount of domain and mathematical expertises(=-=Papalambros & Wilde, 2000-=-). Designers learn from years of experience and significant trialsand error to describe a mathematical model that satisfactorily captures all modelingsrequirements, and guarantees both the existence o... |

89 |
Remembering the past to imagine the future: The prospective brain.
- Schacter, Addis, et al.
- 2007
(Show Context)
Citation Context ...being a constructive process (Schacter & Addis,s2007), and the hypothesis that “the simulation of future episodes is thought to require assystem that can flexibly recombine details from past events” (=-=Schacter et al., 2007-=-, p.659).sThat is, remembering past episodes and constructing future ones both require the retrieval ofsinformation, but more radically, the construction of a future task requires “that event detailss... |

74 |
Expertise in Design: An Overview,”
- Cross
- 2004
(Show Context)
Citation Context ... experiences. A conscious exposition to a large number of examplessand problems from the design domain and accumulation of experience are necessary factorssfor the development of expertise in design (=-=Cross, 2004-=-). As such, it can be claimed thatsdesign experiences, and by implication, design representations, involve, in part, the use ofsepisodic memory and semantic memory. For example, studies on analogy mak... |

51 | Identifying modular and integrative systems and their impact on design team interactions”.
- Sosa, Eppinger, et al.
- 2003
(Show Context)
Citation Context ...5). The general conceptual approach presented as the basis of mosts65 decomposition methods presented in the literature, e.g., (Li & Li, 2005; Michelena &sPapalambros, 1997; Pimmler & Eppinger, 1994; =-=Sosa et al., 2003-=-; Wagner & Papalambros,s1993a) corresponds to Simon’s definition (Simon, 1969/ 1981) of a “nearly decomposablessystem” – a system that is characterized by weakly interacting sub-systems with strong lo... |

43 | A hypergraph framework for optimal model-based decomposition of design problems”.
- Michelena, Papalambros
- 1997
(Show Context)
Citation Context ...sis. In this thesis, a symbolic-mathematical model is considered to be a syntactic encodingsof semantic structural-behavioral design knowledge. Such a model can be stated as a generalsdesign problem (=-=Michelena & Papalambros, 1997-=-), an optimal design problem (Papalambross& Wilde, 2000), a standard non-analytic Functional Dependence Table (FDT) or a DesignsStructure Matrix (DSM) (Li & Li, 2005; Pimmler & Eppinger, 1994). A symb... |

41 | Spontaneous Analogising in Engineering Design: A Comparative Analysis of Experts and Novices,” Des.
- Ball, Ormerod, et al.
- 2004
(Show Context)
Citation Context ...can be claimed thatsdesign experiences, and by implication, design representations, involve, in part, the use ofsepisodic memory and semantic memory. For example, studies on analogy making in designs(=-=Ball et al., 2004-=-) have found that when designers apply abstract experiential knowledge insdesign problems, they use schema-driven analogizing (corresponding to semantic memory),sand when they apply specific prior des... |

41 | A Singularly Valuable Decomposition: The SVD of a Matrix
- Kalman
- 1996
(Show Context)
Citation Context ...claim is that SVD and the correct choice of dimensionality to view these relations aresable to reveal the semantic patterns and the “latent” semantic meaning of the text. In digitalsimage processing (=-=Kalman, 1996-=-; Strang, 2003), SVD is used as a mathematical tool tosidentify pattern redundancy in image data for compression of images. Further, this researchsfound that SVD, as an algorithm, has the capacity to ... |

36 | The Latent Semantic Approach to Studying Design Team Communication. Design Studies, - Dong - 2005 |

31 | Using Modeling Knowledge to Guide Design Space Search. - Gelsey, Schwabacher, et al. - 1998 |

30 | Text Analysis for Constructing Design Representations, - Dong, Agogino - 1997 |

30 |
Robust singular value decomposition analysis of microarray data.
- Liu, Hawkins, et al.
- 2003
(Show Context)
Citation Context ... (Dong,s2005), prediction of psychological phenomenon (Wolfe & Goldman, 2003), digital imagesprocessing (Kalman, 1996), internet search algorithms (Strang, 2003), and clustering genesmicroarray data (=-=Liu et al., 2003-=-). In all these approaches, the underlying “common”srepresentation on which SVD operates is a matrix that captures relationships between one orstwo types of “things” or “concepts”, commonly one type s... |

25 |
Artificial Intelligence: An Empirical Science.
- Simon
- 1995
(Show Context)
Citation Context ...velop computational methods that could assist the designer with problem modeling andsreformulation tasks.s5 1.1.2 Design theory motivationsThe physical symbol system hypothesis (Newell & Simon, 1976; =-=Simon, 1995-=-) establishes asprevalent methodological and theoretical approach for developing AI algorithms. This is thesexplicit symbolic reasoning approach (Cagan et al., 1997), which specifies that knowledge is... |

22 |
A General Framework For Decomposition Analysis
- Wagner, Papalambros
- 1993
(Show Context)
Citation Context ...arameters or system components. In some cases, the mappingsdefined can also be other than Boolean. In general, non-analytical models can either bes46 described as a Functional Dependence Table (FDT) (=-=Wagner & Papalambros, 1993-=-a), or asDesign Structure Matrix (DSM) form (Pimmler & Eppinger, 1994). Other forms of nonanalytical representations, like the Multiple Domain Matrix (Lindemann et al., 2009) exist.sThough this thesis... |

20 | The A-Design Approach to Managing Automated Design Synthesis,” Res. - Campbell, Cagan, et al. - 2003 |

20 | Decision-based collaborative optimization” - Gu, Renaud, et al. - 2002 |

20 | The fundamental theorem of linear algebra - Strang - 1993 |

18 |
LAPACK users’ guide, 3rd edition
- Anderson, Bai, et al.
- 1999
(Show Context)
Citation Context ...n time for the full method will be dominated by the SVD and K-meanssclustering algorithms. In practice, fast and efficient algorithms exist for computing both thesSVD and K-means. MATLAB uses LAPACK (=-=Anderson, 1999-=-). In design modeling, theslargest matrices would go to thousands of rows and columns. The time complexityscalculations only become important for matrices of dimension > 105. REIFORM has not beenstest... |

16 |
Decomposition and Representation Methods in Mechanical Design,’’
- Kusiak, Larson
- 1995
(Show Context)
Citation Context ...97) and Wagners(1993a)).sDecomposition approaches have been classified in various ways – object-based, aspectbased or sequential (Michelena & Papalambros, 1997) or product, problem and process baseds(=-=Kusiak & Larson, 1995-=-). The general conceptual approach presented as the basis of mosts65 decomposition methods presented in the literature, e.g., (Li & Li, 2005; Michelena &sPapalambros, 1997; Pimmler & Eppinger, 1994; S... |

15 | A transformation system for interactive reformulation of design optimization strategies
- Ellman, Keane, et al.
- 1998
(Show Context)
Citation Context ...he acquired knowledge can be used in the same designsexperience to reformulate the problem.sIn design, the same problem can be formulated and solved in a number of different wayss(Cagan et al., 1997; =-=Ellman et al., 1998-=-). It is common for design problems to have differingsformulations, even if they belong to the same design domain. Two designers may choose tosmodel the same problem in different ways. Often, the spec... |

14 | SYMON: Automated symbolic monotonicity analysis system for qualitative design optimization. - Choy, Agogino - 1986 |

12 |
The ‘What’ and ‘How’ of Learning in Design
- DUFFY
- 1997
(Show Context)
Citation Context ...w of existing approaches: establishing the first three behaviorscriteria for the methodsThe application of AI and learning algorithms to design problems is a wide area of research.sExtensive reviews (=-=Duffy, 1997-=-; Grecu & Brown, 1998; Sim & Duffy, 1998) show that mostsresearch endeavors focus on developing specific knowledge acquisition and inferencesmethods for specific design domains, representational forms... |

12 | Learning to set up numerical optimizations of engineering designs - Schwabacher, Ellman, et al. - 1998 |

12 | Activity analysis: The qualitative analysis of stationary points for optimal reasoning
- Williams, Cagan
- 1994
(Show Context)
Citation Context ...roblem was chosen as a demonstration problem because it isswidely referred to in the literature and serves as a test problem for many approachess(Michelena & Agogino, 1988; Papalambros & Wilde, 2000; =-=Williams & Cagan, 1994-=-). It hassbeen formulated and solved in multiple ways. This thesis works with the commonly knowns105 formulation of the problem – a single objective formulation (Papalambros & Wilde, 2000).sFigure 7.1... |

11 |
System Integration Analysis Of A Large Commercial Aircraft Engine”,
- Rowles
- 1999
(Show Context)
Citation Context ....s78 5.2.1 Example demonstrationsModularity analysis is demonstrated on a large scale design problem. Figure 5.9 shows thesDSM representation for a large Pratt and Whitney commercial aircraft engine (=-=Rowles, 1999-=-;sSosa et al., 2003) with 54 design components in 8 subsystems (Fan, Low PressuresCompressor (LPC), High Pressure Compressor (HPC), Combustion Chamber (CC), HighsPressure Turbine (HPT), Low Pressure T... |

10 | Analysis of decomposability and complexity for design problems in the context of decomposition
- Chen, Li
- 2005
(Show Context)
Citation Context ...alsdesign problem (Michelena & Papalambros, 1997), an optimal design problem (Papalambross& Wilde, 2000), a standard non-analytic Functional Dependence Table (FDT) or a DesignsStructure Matrix (DSM) (=-=Li & Li, 2005-=-; Pimmler & Eppinger, 1994). A symbol in a designsmodel represents a structure or behavior variable (for e.g. internal diameter, force), parameters(e.g. minimum wall thickness), or system component (e... |

10 | Constructive memory: the ghosts of past and future." Nature 445(7123 - Schacter, Addis - 2007 |

9 |
Multi-Objective Hydraulic Cylinder Design,"
- Michelena, Agogino
- 1988
(Show Context)
Citation Context ...ylinder design problemsThe hydraulic cylinder design problem was chosen as a demonstration problem because it isswidely referred to in the literature and serves as a test problem for many approachess(=-=Michelena & Agogino, 1988-=-; Papalambros & Wilde, 2000; Williams & Cagan, 1994). It hassbeen formulated and solved in multiple ways. This thesis works with the commonly knowns105 formulation of the problem – a single objective ... |

8 | The use of artificial intelligence to improve the numerical optimization of complex engineering designs - Schwabacher - 1996 |

7 |
Implementation of decomposition analysis in optimal design
- Wagner, Papalambros
- 1993
(Show Context)
Citation Context ...arameters or system components. In some cases, the mappingsdefined can also be other than Boolean. In general, non-analytical models can either bes46 described as a Functional Dependence Table (FDT) (=-=Wagner & Papalambros, 1993-=-a), or asDesign Structure Matrix (DSM) form (Pimmler & Eppinger, 1994). Other forms of nonanalytical representations, like the Multiple Domain Matrix (Lindemann et al., 2009) exist.sThough this thesis... |

5 |
A foundation for machine learning in design
- Sim, Duffy
- 1998
(Show Context)
Citation Context ...hing the first three behaviorscriteria for the methodsThe application of AI and learning algorithms to design problems is a wide area of research.sExtensive reviews (Duffy, 1997; Grecu & Brown, 1998; =-=Sim & Duffy, 1998-=-) show that mostsresearch endeavors focus on developing specific knowledge acquisition and inferencesmethods for specific design domains, representational forms or for specific design tasks. Thissis u... |

4 |
A conceptual framework for combining artificial intelligence and optimization in engineering design
- Cagan, Grossman, et al.
- 1997
(Show Context)
Citation Context ...l system hypothesis (Newell & Simon, 1976; Simon, 1995) establishes asprevalent methodological and theoretical approach for developing AI algorithms. This is thesexplicit symbolic reasoning approach (=-=Cagan et al., 1997-=-), which specifies that knowledge issexplicitly represented in terms of symbols and symbolic processes. This theory has had asmajor influence on the development and application of AI algorithms in des... |

3 |
Dimensions of machine learning
- Grecu, Brown
- 1998
(Show Context)
Citation Context ... approaches: establishing the first three behaviorscriteria for the methodsThe application of AI and learning algorithms to design problems is a wide area of research.sExtensive reviews (Duffy, 1997; =-=Grecu & Brown, 1998-=-; Sim & Duffy, 1998) show that mostsresearch endeavors focus on developing specific knowledge acquisition and inferencesmethods for specific design domains, representational forms or for specific desi... |

3 | Cognitive investigations into knowledge representation in engineering design - Moss, Kotovsky, et al. - 2004 |

2 | Conceptual Coordination - Clancey - 1999 |

1 |
Modeling the multiple-domain matrix
- Lindemann, Maurer, et al.
- 2009
(Show Context)
Citation Context ...ependence Table (FDT) (Wagner & Papalambros, 1993a), or asDesign Structure Matrix (DSM) form (Pimmler & Eppinger, 1994). Other forms of nonanalytical representations, like the Multiple Domain Matrix (=-=Lindemann et al., 2009-=-) exist.sThough this thesis does not demonstrate method application on examples from all the possiblesnon-analytical representations, it is easy to see that the method is applicable on anysrepresentat... |

1 | Learning from design experience in an agent based system - Moss, Kotovsky, et al. - 2004 |

1 | Artificial Intelligence: A modern approach - thesis - 2003 |