機械学習:Machine Learning

Protected: Application of Topic Models to Information Other Than Documents – Application to Image Data and Graph Data (Stochastic Block Model, Mixed Member Probabilistic Block Model)

Topic models for image and graph data using stochastic block models for digital transformation and artificial intelligence tasks.

2021.12.24

微分積分:Calculus推論技術:inference Technology最適化:Optimization機械学習:Machine Learning確率・統計:Probability and Statistics自然言語処理:Natural Language Processing

MCMC and Bayesian estimation

デジタルトランスフォーメーション(DX)、人工知能(AI)タスクに活用される確率関数の積分等に用いられるマルコフ連鎖モンテカルロ

2021.12.23

推論技術:inference Technology最適化:Optimization機械学習:Machine Learning確率・統計:Probability and Statistics

Protected: Extension of topic models (adding structure to topics) Correlation topic model, slingshot distribution model with hierarchical structure, probabilistic latent semantic visualization with low-dimensional spatial structure

Overview of topic models with structure in correlated topics used in digital transformation and artificial intelligence tasks (correlated topic model, slingshot distribution model with hierarchical structure, probabilistic latent meaning visualization with low-dimensional spatial structure)

2021.12.23

微分積分:Calculus推論技術:inference Technology最適化:Optimization機械学習:Machine Learning確率・統計:Probability and Statistics自然言語処理:Natural Language Processing

Protected: Extending topic models (using other information as well) (2) Noisy correspondence topic model, author topic model, topic tracking model

Among topic models that rely on auxiliary information for digital transformation and artificial intelligence tasks, we will discuss noisy topic models, author topic models, and topic tracking models.

2021.12.22

微分積分:Calculus推論技術:inference Technology最適化:Optimization機械学習:Machine Learning確率・統計:Probability and Statistics自然言語処理:Natural Language Processing

Protected: Extending the topic model (using other information) (1) Combined topic model and corresponding topic model

Create a topic model with auxiliary information to be used for digital transformation and artificial intelligence tasksJoining / Corresponding Topic Model Overview

2021.12.21

推論技術:inference Technology最適化:Optimization機械学習:Machine Learning確率・統計:Probability and Statistics自然言語処理:Natural Language Processing

Topic Model Theory and Implementation

A topic model is a probability generation model for extracting topics from sentences, which is one of the natural language processing technologies used in digital transformation and artificial intelligence tasks.

2021.12.17

機械学習:Machine Learning確率・統計:Probability and Statistics自然言語処理:Natural Language Processing

Protected: Change detection by density ratio estimation – Detection of structural changes using the Kullback-Leibler density ratio estimation method

Detecting structural changes using the Kullback-Leibler density ratio estimation method for digital transformation and artificial intelligence tasks.

2021.12.17

微分積分:Calculus推論技術:inference Technology最適化:Optimization機械学習:Machine Learning異常検知・変化検知確率・統計:Probability and Statistics

Protected: Anomaly detection by density ratio estimation – Anomaly Estimation from Unsupervised Data Using the Kullback-Leibler Density Ratio Estimation Method

Among the anomaly/change detection techniques used in digital transformation and artificial intelligence tasks, I will introduce a method for anomaly detection using probability density ratio for unsupervised data ,Kullback-Leibler density ratio estimation method

2021.12.16

微分積分:Calculus推論技術:inference Technology最適化:Optimization機械学習:Machine Learning異常検知・変化検知確率・統計:Probability and Statistics

Protected: Anomaly detection using sparse structure learning- Graph models and regularization that link broken dependencies between variables to anomalies.

Graph models and regularization that link broken dependencies between variables to anomalies.

2021.12.15

微分積分:Calculus推論技術:inference Technology最適化:Optimization機械学習:Machine Learning異常検知・変化検知確率・統計:Probability and Statistics

Protected: Change detection using subspace method -Singular spectral transform method for time series data

Singular spectral transform (SVD) method for extracting change points from time series data for digital transformation and artificial intelligence tasks.

2021.12.14

微分積分:Calculus推論技術:inference Technology最適化:Optimization機械学習:Machine Learning異常検知・変化検知確率・統計:Probability and Statistics