| MOHALI MARCANO ANDRES MANUEL | M, 2回目発表 | 数理情報学 | 池田 和司 | 松本 健一 | 久保 孝富 | 日永田 智絵 | Li Yuzhe |
|
title: Behavioral Analysis of Mice with Disomy Deletion of Chromosome 15: Base for Pharmacological Approach
abstract: Disomy deletion of chromosome 15 in mice has been linked to metabolic, endocrine, and psychiatric disorders, notably autism spectrum disorders. Previous studies have suggested short-term social behavior differences in these mice, but long-term social interactions remain largely unexplored. This study aims to analyze the long-term social behavior of mice with chromosome 15 disomy deletion using DeepLabCut (DLC) for pose estimation and neural network models for behavioral analysis. By leveraging the interpretability of embeddings, we project behavioral tokens into a reduced vector space to assess differences in social interactions. The experiment involves tracking specific body parts of control and sick mice in various social settings, constructing social interaction tokens, and using LSTM neural networks to construct the next token prediction and autoencoders to analyze these tokens. Our approach aims to provide insights into the behavioral effects of chromosome 15 disomy deletion, useful to explore potential genetic pharmacology and optogenetics interventions to restore normal behavior. Preliminary results indicate distinct differences in the embedding projections of nose-to-nose interactions and investigative behaviors. These differences were assessed by examining the variance in cosine similarity and by analyzing the differences in sequence reconstruction when using opposite models. The analysis revealed variations in specific transitions within the k-mer sequences. language of the presentation: English | |||||||
| YU ZIHAO | D, 中間発表 | 数理情報学 | 池田 和司 | 松本 健一 | 久保 孝富 | 日永田 智絵 | Li Yuzhe |
|
title: Bus Routing from Taxi Data
abstract: The study analyzes origin-destination data from taxi operations and seeks to propose optimal bus routes. First, we establish the effectiveness of hierarchical density-based spatial clustering of applications with noise in determining bus stop locations. An original decision-diagram-based algorithm then determines the bus routes through a reward-penalty procedure. To close, we evaluate the coverage of the resulting bus routes against the given taxi data. language of the presentation: English | |||||||