Limits...
Using Graph Components Derived from an Associative Concept Dictionary to Predict fMRI Neural Activation Patterns that Represent the Meaning of Nouns.

Akama H, Miyake M, Jung J, Murphy B - PLoS ONE (2015)

Bottom Line: We measure the effectiveness of graph-based coefficients through the application of linguistic graph information for a neural activity recorded during conceptual processing in the human brain.Furthermore, correlating the voxel information with the MiF-based principal components, a new computational neurolinguistics model with a network connectivity paradigm is created.This allows two dimensions of context space to be incorporated with both semantic and neural distributional representations.

View Article: PubMed Central - PubMed

Affiliation: Graduate School of Decision Science and Technology, Tokyo Institute of Technology, Tokyo, Japan.

ABSTRACT
In this study, we introduce an original distance definition for graphs, called the Markov-inverse-F measure (MiF). This measure enables the integration of classical graph theory indices with new knowledge pertaining to structural feature extraction from semantic networks. MiF improves the conventional Jaccard and/or Simpson indices, and reconciles both the geodesic information (random walk) and co-occurrence adjustment (degree balance and distribution). We measure the effectiveness of graph-based coefficients through the application of linguistic graph information for a neural activity recorded during conceptual processing in the human brain. Specifically, the MiF distance is computed between each of the nouns used in a previous neural experiment and each of the in-between words in a subgraph derived from the Edinburgh Word Association Thesaurus of English. From the MiF-based information matrix, a machine learning model can accurately obtain a scalar parameter that specifies the degree to which each voxel in (the MRI image of) the brain is activated by each word or each principal component of the intermediate semantic features. Furthermore, correlating the voxel information with the MiF-based principal components, a new computational neurolinguistics model with a network connectivity paradigm is created. This allows two dimensions of context space to be incorporated with both semantic and neural distributional representations.

No MeSH data available.


Friendship network of Zachary’s famous “Karate Club”.There is one shortest path between vertices 2 and 7 (red edges), with a step length of 2. It follows from the sum of the elements in the second and seventh rows (or columns) of the second power of the adjacency matrix that there are 52 and 25 two-step paths starting from vertices 2 and 7, respectively. Thus, the Jaccard similarity between them is calculated as (52+25)−1 = 0.012987, if we take into account all of the steps starting from each of the two vertices that have a step length of 2. In this figure, the yellow nodes are reachable in two steps from both vertices 2 and 7, whereas, under the same path condition, the blue nodes can only be reached from vertex 2, and the green node can only be reached from vertex 7. In addition, these two vertices have a Simpson coefficient of 25−1 = 0.04 and a MiF value of 0.0185583. It is widely known that the friendship network among the Karate club members was split into two factions. According to the degree to which the final attachments to each faction match with the results of graph clustering, it is possible to evaluate the effectiveness of the clustering technique (based on an adjacency matrix) for simulating the social relationships. The two factions are represented here by the vertex labels with red italic font (one group composed of vertices {1, 2, 3, 4, 5, 6, 7, 8, 11, 12, 13, 14, 17, 18, 20, 22}) and those with blue bold font (the other group of {9, 10, 15, 16, 19, 21, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34}). Misclassification always occurred by binding vertices 3 and 10 at early stages when the Jaccard index, Simpson index, and MiF with the default β value (0.5) were applied to the hierarchical graph clustering of this network. With a small value of β (for example, 0.01), which can reflect the asymmetrical roles played by the two agents in terms of connectivity, MiF predicts the composition of the two factions with 100% accuracy. For further details, see S2 Program. This figure was created using Mathematica 8.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4482269&req=5

pone.0125725.g001: Friendship network of Zachary’s famous “Karate Club”.There is one shortest path between vertices 2 and 7 (red edges), with a step length of 2. It follows from the sum of the elements in the second and seventh rows (or columns) of the second power of the adjacency matrix that there are 52 and 25 two-step paths starting from vertices 2 and 7, respectively. Thus, the Jaccard similarity between them is calculated as (52+25)−1 = 0.012987, if we take into account all of the steps starting from each of the two vertices that have a step length of 2. In this figure, the yellow nodes are reachable in two steps from both vertices 2 and 7, whereas, under the same path condition, the blue nodes can only be reached from vertex 2, and the green node can only be reached from vertex 7. In addition, these two vertices have a Simpson coefficient of 25−1 = 0.04 and a MiF value of 0.0185583. It is widely known that the friendship network among the Karate club members was split into two factions. According to the degree to which the final attachments to each faction match with the results of graph clustering, it is possible to evaluate the effectiveness of the clustering technique (based on an adjacency matrix) for simulating the social relationships. The two factions are represented here by the vertex labels with red italic font (one group composed of vertices {1, 2, 3, 4, 5, 6, 7, 8, 11, 12, 13, 14, 17, 18, 20, 22}) and those with blue bold font (the other group of {9, 10, 15, 16, 19, 21, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34}). Misclassification always occurred by binding vertices 3 and 10 at early stages when the Jaccard index, Simpson index, and MiF with the default β value (0.5) were applied to the hierarchical graph clustering of this network. With a small value of β (for example, 0.01), which can reflect the asymmetrical roles played by the two agents in terms of connectivity, MiF predicts the composition of the two factions with 100% accuracy. For further details, see S2 Program. This figure was created using Mathematica 8.

Mentions: To give the co-occurrence adjustment, it is known that the Jaccard similarity can be intuitively formulated as/A∩B//A∪B/(1)for two sets A and B. Indeed, for two vertices, this index is usually computed as/N(a)∩N(b)//N(a)∪N(b)/,(1)'where N(a) denotes the set of all neighbours of vertex a. To enhance the accuracy with which the distance between remote nodes is evaluated, we extend the interpretation of expression (1) such that the numerator is the distance of the shortest path connecting vertices a and b. The denominator in (1) is the sum of the degrees of vertices a and b, or, in some cases, all of the steps starting from these vertices that have an identical step length. In this article, we adopt the latter definition for the denominator, and set the step length equal to the shortest path between a and b in the numerator. Fig 1 illustrates this coefficient using the friendship network of Zachary’s famous “Karate Club” [44].


Using Graph Components Derived from an Associative Concept Dictionary to Predict fMRI Neural Activation Patterns that Represent the Meaning of Nouns.

Akama H, Miyake M, Jung J, Murphy B - PLoS ONE (2015)

Friendship network of Zachary’s famous “Karate Club”.There is one shortest path between vertices 2 and 7 (red edges), with a step length of 2. It follows from the sum of the elements in the second and seventh rows (or columns) of the second power of the adjacency matrix that there are 52 and 25 two-step paths starting from vertices 2 and 7, respectively. Thus, the Jaccard similarity between them is calculated as (52+25)−1 = 0.012987, if we take into account all of the steps starting from each of the two vertices that have a step length of 2. In this figure, the yellow nodes are reachable in two steps from both vertices 2 and 7, whereas, under the same path condition, the blue nodes can only be reached from vertex 2, and the green node can only be reached from vertex 7. In addition, these two vertices have a Simpson coefficient of 25−1 = 0.04 and a MiF value of 0.0185583. It is widely known that the friendship network among the Karate club members was split into two factions. According to the degree to which the final attachments to each faction match with the results of graph clustering, it is possible to evaluate the effectiveness of the clustering technique (based on an adjacency matrix) for simulating the social relationships. The two factions are represented here by the vertex labels with red italic font (one group composed of vertices {1, 2, 3, 4, 5, 6, 7, 8, 11, 12, 13, 14, 17, 18, 20, 22}) and those with blue bold font (the other group of {9, 10, 15, 16, 19, 21, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34}). Misclassification always occurred by binding vertices 3 and 10 at early stages when the Jaccard index, Simpson index, and MiF with the default β value (0.5) were applied to the hierarchical graph clustering of this network. With a small value of β (for example, 0.01), which can reflect the asymmetrical roles played by the two agents in terms of connectivity, MiF predicts the composition of the two factions with 100% accuracy. For further details, see S2 Program. This figure was created using Mathematica 8.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4482269&req=5

pone.0125725.g001: Friendship network of Zachary’s famous “Karate Club”.There is one shortest path between vertices 2 and 7 (red edges), with a step length of 2. It follows from the sum of the elements in the second and seventh rows (or columns) of the second power of the adjacency matrix that there are 52 and 25 two-step paths starting from vertices 2 and 7, respectively. Thus, the Jaccard similarity between them is calculated as (52+25)−1 = 0.012987, if we take into account all of the steps starting from each of the two vertices that have a step length of 2. In this figure, the yellow nodes are reachable in two steps from both vertices 2 and 7, whereas, under the same path condition, the blue nodes can only be reached from vertex 2, and the green node can only be reached from vertex 7. In addition, these two vertices have a Simpson coefficient of 25−1 = 0.04 and a MiF value of 0.0185583. It is widely known that the friendship network among the Karate club members was split into two factions. According to the degree to which the final attachments to each faction match with the results of graph clustering, it is possible to evaluate the effectiveness of the clustering technique (based on an adjacency matrix) for simulating the social relationships. The two factions are represented here by the vertex labels with red italic font (one group composed of vertices {1, 2, 3, 4, 5, 6, 7, 8, 11, 12, 13, 14, 17, 18, 20, 22}) and those with blue bold font (the other group of {9, 10, 15, 16, 19, 21, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34}). Misclassification always occurred by binding vertices 3 and 10 at early stages when the Jaccard index, Simpson index, and MiF with the default β value (0.5) were applied to the hierarchical graph clustering of this network. With a small value of β (for example, 0.01), which can reflect the asymmetrical roles played by the two agents in terms of connectivity, MiF predicts the composition of the two factions with 100% accuracy. For further details, see S2 Program. This figure was created using Mathematica 8.
Mentions: To give the co-occurrence adjustment, it is known that the Jaccard similarity can be intuitively formulated as/A∩B//A∪B/(1)for two sets A and B. Indeed, for two vertices, this index is usually computed as/N(a)∩N(b)//N(a)∪N(b)/,(1)'where N(a) denotes the set of all neighbours of vertex a. To enhance the accuracy with which the distance between remote nodes is evaluated, we extend the interpretation of expression (1) such that the numerator is the distance of the shortest path connecting vertices a and b. The denominator in (1) is the sum of the degrees of vertices a and b, or, in some cases, all of the steps starting from these vertices that have an identical step length. In this article, we adopt the latter definition for the denominator, and set the step length equal to the shortest path between a and b in the numerator. Fig 1 illustrates this coefficient using the friendship network of Zachary’s famous “Karate Club” [44].

Bottom Line: We measure the effectiveness of graph-based coefficients through the application of linguistic graph information for a neural activity recorded during conceptual processing in the human brain.Furthermore, correlating the voxel information with the MiF-based principal components, a new computational neurolinguistics model with a network connectivity paradigm is created.This allows two dimensions of context space to be incorporated with both semantic and neural distributional representations.

View Article: PubMed Central - PubMed

Affiliation: Graduate School of Decision Science and Technology, Tokyo Institute of Technology, Tokyo, Japan.

ABSTRACT
In this study, we introduce an original distance definition for graphs, called the Markov-inverse-F measure (MiF). This measure enables the integration of classical graph theory indices with new knowledge pertaining to structural feature extraction from semantic networks. MiF improves the conventional Jaccard and/or Simpson indices, and reconciles both the geodesic information (random walk) and co-occurrence adjustment (degree balance and distribution). We measure the effectiveness of graph-based coefficients through the application of linguistic graph information for a neural activity recorded during conceptual processing in the human brain. Specifically, the MiF distance is computed between each of the nouns used in a previous neural experiment and each of the in-between words in a subgraph derived from the Edinburgh Word Association Thesaurus of English. From the MiF-based information matrix, a machine learning model can accurately obtain a scalar parameter that specifies the degree to which each voxel in (the MRI image of) the brain is activated by each word or each principal component of the intermediate semantic features. Furthermore, correlating the voxel information with the MiF-based principal components, a new computational neurolinguistics model with a network connectivity paradigm is created. This allows two dimensions of context space to be incorporated with both semantic and neural distributional representations.

No MeSH data available.