Mathematica（Wolfram言語）V12.1および近日リリース予定のV12.2における新機能，改良点のいくつかを紹介します．特に，ニューラルネットワークの応用として注目を集めるGAN（敵対的生成ネットワーク）や，ビデオ処理機能，さらに近年のコンファレンスで取り上げた偏微分方程式の数値求解，最適化計算についての新しいトピックについてお話しする予定です．

Many SEIR-type epidemiology models for COVID-19 have been submitted to preprint servers since the beginning of the year, and some have been published. Nearly all these models have been based on deterministic differential equation models, which assume large, rapidly and well-mixed populations. These assumptions rarely hold for real biological populations, and other methods may be more appropriate. One frequent feature of these "published" models and observed in our own models is the need to use artificially small total population sizes to fit the reported number of cases and deaths in a given country or region. Network-based models offer an alternative to the well-mixed assumption, and stochastic methods provide an alternative to the rapid-mixing assumption. Preliminary results combining our previously developed stochastic simulation method using the Gillespie algorithm with a network architecture of subpopulations have been encouraging. In this talk, we will describe our ongoing efforts to fully explore the effects of network topology and various mixing scenarios on the dynamics of disease spread throughout the entire population. Comparison of the modeled outcomes with some well-characterized COVID-19 data from Europe and China will show whether or not this approach is successful, or whether other modeling strategies are needed.

人工透析領域の日常診療や臨床研究においてMathematicaはkinetic modelingや複雑な数式計算や複合データの臨床分類などに利用されてきた．さらに近年，機械学習による治療効率予測への応用も始まっている．本発表では，これらの解析をMathematicaならではの視覚化機能を生かしながら紹介したい．

Molecule fingerprints are a method of representing a molecule as a sequence of bits, either on or off, that still encodes important information about the molecular structure. These fingerprints can be used for a "molecular distance function" for substructure screening or as inputs to machine learning functions. This talk will introduce the concept of molecule fingerprints and show the different fingerprint types coming in Version 12.2 of the Wolfram Language, and showcase many enhancements to MoleculePlot.

対面授業とオンライン授業を織り交ぜたハイブリッドな教育スタイルへと変化する中で，Wolfram|Alpha Pro，Wolfram|Alpha Notebook Edition，Wolfram Programming Lab，そしてMathematica | Online等，Wolframのクラウドサービスが，ニューノーマル時代の教育・学習ツールとしてどのように役に立つのかを事例を交えて紹介します．

A quick tour of a few fun side projects I have made including Pi Slices Daily, a COVID dashboard and a baseball-player network story creator. I will show you how I combine the cloud, the Wolfram Function Repository, notebooks and Mathematica to turn ideas into real live things (even if they are hacky).

シミュレーションによるソフトウェアメトリックス最適値の評価，開発プロセスにおける問題点の発見，シナリオベーステストケース生成について説明します．また，ニューラルネットの一種であるAutoEncoderを，自動車に対するサイバー攻撃検出に利用した例も紹介します．

Spatial point data, also known as spatial point patterns, refers to collections of points (or events) in space. Examples include trees in a forest, gold deposits, positions of stars, earthquakes, crime locations, animal sightings, etc. Spatial data analysis, as a statistical exploration of point patterns, aims to answer questions about spatial randomness, point intensity and interpoint interactions. In this talk, we will explore the existing and upcoming Wolfram Language functionality designed for the analysis and modeling of spatial point data.

Wolfram Application Server (WAS) is a new clustered solution for providing scalable, highly available access to Wolfram Language APIs, forms and other active web elements. It provides a rapid deployment environment for Wolfram Language applications that can grow to meet the full needs of a production environment. In this talk, we will discuss the overall architecture and capabilities of WAS and demonstrate how developers can deploy content to the server making it accessible to end users.