A neural network-based machine learning approach for supporting synthesis

AbstractThe goal of machine learning for artifact synthesis is the acquisition of the relationships among form, function, and behavior properties that can be used to determine more directly form attributes that satisfy design requirements. The proposed approach to synthesis knowledge acquisition and use (SKAU) described in this paper, called NETSYN, creates a function to estimate the probability of each possible value of each design property being used in a given design context. NETSYN uses a connectionist learning approach to acquire and represent this probability estimation function and exhibits good performance when tested on an artificial design problem. This paper presents the NETSYN approach for SKAU, a preliminary test of its capability, and a discussion of issues that need to be addressed in future work.

Download Full-text

A machine learning approach to product review disambiguation based on function, form and behavior classification

Decision Support Systems ◽

10.1016/j.dss.2017.03.007 ◽

2017 ◽

Vol 97 ◽

pp. 81-91 ◽

Cited By ~ 16

Author(s):

Abhinav Singh ◽

Conrad S. Tucker

Keyword(s):

Machine Learning ◽

Learning Approach ◽

Function Form ◽

Product Review ◽

Machine Learning Approach ◽

And Behavior

Download Full-text

Opponent Provocation and Behavior Classification: A Machine Learning Approach

RoboCup 2007: Robot Soccer World Cup XI - Lecture Notes in Computer Science ◽

10.1007/978-3-540-68847-1_58 ◽

2008 ◽

pp. 540-547 ◽

Cited By ~ 5

Author(s):

Ramin Fathzadeh ◽

Vahid Mokhtari ◽

Mohammad Reza Kangavari

Keyword(s):

Machine Learning ◽

Learning Approach ◽

Machine Learning Approach ◽

And Behavior

Download Full-text

Outomatiese lemma-identifisering vir Afrikaans

Literator ◽

10.4102/lit.v29i1.101 ◽

2008 ◽

Vol 29 (1) ◽

pp. 65-92

Author(s):

H.J. Groenewald ◽

G.B. Van Huyssteen

Keyword(s):

Machine Learning ◽

Feature Selection ◽

Text Processing ◽

Learning Approach ◽

Input And Output ◽

Base Form ◽

Alternative Approach ◽

Machine Learning Approach ◽

Word Forms ◽

Future Work

Automatic lemmatisation for Afrikaans Automatic lemmatisation is a general normalisation procedure in text processing, where all inflected forms of a lexical word are normalised to a single lemma (i.e. a meaningful, uninflected base form from which more complex word forms could be formed). Traditionally, lemmatisers are developed by writing language-specific rules to identify lemmas. In this article an alternative approach is investigated, namely a machine learning approach, to develop a lemmatiser for Afrikaans (LIA: “Lemmaidentifiseerder vir Afrikaans”). An overview regarding the process of inflection in Afrikaans is provided with the aim of identifying the categories of inflection that are relevant for lemmatisation in Afrikaans. The format of the input and output is described with special reference to the nine inflectional categories for Afrikaans that the system should be able to handle. Then the task of lemmatisation as a classification task for machine learning is described, and a concise introduction to memory-based learning is provided. The development and evaluation of LIA is discussed in detail, and it is illustrated how the performance of the initial classifier is improved through feature selection and parameter optimisation. The best classifier reaches an accuracy of 92,8%. The article concludes with a view on some future work.

Download Full-text

Unsupervised Machine Learning Approach for Identifying Biomechanical Influences on Protein-Ligand Binding Affinity

10.21203/rs.3.rs-919250/v1 ◽

2021 ◽

Author(s):

Arjun Singh

Keyword(s):

Machine Learning ◽

Binding Affinity ◽

Principal Component ◽

3D Models ◽

Supervised Machine Learning ◽

Learning Approach ◽

Drug Candidates ◽

Machine Learning Approach ◽

Novel Drug ◽

Future Work

Abstract Drug discovery is incredibly time-consuming and expensive, averaging over 10 years and $985 million per drug. Calculating the binding affinity between a target protein and a ligand is critical for discovering viable drugs. Although supervised machine learning (ML) can predict binding affinity accurately, models experience severe overfitting due to an inability to identify informative properties of protein-ligand complexes. This study used unsupervised ML to reveal underlying protein-ligand characteristics that strongly influence binding affinity. Protein-ligand 3D models were collected from the PDBBind database and vectorized into 2422 features per complex. Principal Component Analysis (PCA), t-Distributed Stochastic Neighbor Embedding (t-SNE), K-Means Clustering, and heatmaps were used to identify groups of complexes and the features responsible for the separation. ML benchmarking was used to determine the features’ effect on ML performance. The PCA heatmap revealed groups of complexes with binding affinity of pKd < 6 and pKd > 8, and identified the number of CCCH and CCCCCH fragments in the ligand as the most responsible features. A high correlation of 0.8337, their ability to explain 18% of the binding affinity’s variance, and an error increase of 0.09 in Decision Trees when trained without the two features suggests that the fragments exist within a larger ligand substructure that significantly influences binding affinity. This discovery is a baseline for informative ligand representations to be generated so that ML models overfit less and can more reliably identify novel drug candidates. Future work will focus on validating the ligand substructure’s presence and discovering more informative intra-ligand relationships.

Download Full-text