1. Lecture: Recurrent neural network and their application;
Brief lecture outline:
During lecture feed forward neural and recurent neural networks, as well as LSTM networks will be discussed. Application of RNNs to natural language processing (sentiment analysis; question and answering), text generation. Application to time series will be covered as well.
Working programming language will be Python.
Lecture pre-requisites:
Familiarity with mathematics, probability theory, statistics, and algorithms is expected, on the level it is typically introduced at the bachelor level in computer science or engineering programs.
2. Lecture: Time Series Classification and Clustering
Brief lecture outline:
In this course an overview of machine learning techniques to deal with time series data will be presented. Participants will first consider unsupervised learning methods such as distance-based clustering and feature-based clustering. Clustering is often used to detect typical behaviour in a timeseries. Second, focus on supervised learning techniques where discriminative periods or features of a time series are learned. This can be used to detect what is generating the series. For example, to detect which device is using resources, who is using a phone or whether an unexpected event has occured. Last, participants will briefly look to semi-supervised learning. This is a popular setting in time series monitoring where only part of the series is labeled – this is especially relevant to big data sets.
Working programming language will be Python.
Lecture pre-requisites:
Familiarity with mathematics, probability theory, statistics, and algorithms is expected, on the level it is typically introduced at the bachelor level in computer science or engineering programs.
3. Lecture: Big Data Science and Artificial Intelligence in Banking and Finance
Brief lecture outline:
An important observation regarding Advanced Analytics problems occurring in the field of Banking and Finance is that those cover almost every type of granularity regarding the respective Artificial Intelligence modeling approach. From finding optimal routes to refill ATMs to the design of Predictive Maintenance schedules of these ATMs via Private Banking Product Recommendation Systems and Anti Money Laundering as well as Fraud Detection strategies to high-frequency tick-data trading models – the whole spectrum of modeling approaches is required both from a data perspective as well as a decision perspective. A balanced mix of Simulation, Optimization and Machine Learning is necessary to cope with the underlying problems successfully. Artificial Intelligence can be put on top of these solution techniques to e.g. allow for an automatic re-calibration of the respective decision model to adapt to changing markets or macroeconomic regimes autonomously.
This lecture consists of four parts. Part One contains a lecture-style overview of the above-mentioned range of applications and how to deal with those different aspects of Big Data. The following three parts are delivered as hands-on sessions using predefined Jupyter notebooks that will be handed out to the participants beforehand. Part Two is dedicated to large-scale optimization models for contemporary Robo Advisory asset allocations. Part Three considers Anti-Money Laundering and Fraud Detection strategies using Machine Learning while Part Four is considering Deep Learning approaches to optimally trade assets from different asset classes.
Working programming language will be Python.
Lecture pre-requisites:
Familiarity with mathematics, probability theory, statistics, and algorithms is expected, on the level it is typically introduced at the bachelor level in computer science or engineering programs.