Fintech (INstitutions) Flashcards

Question

What is non-supervised learning?

Answer 1

Training models on unlabelled data to form clusters based on similar properties

Answer 2

A type of ML where the model learns an optimal policy to maximize a reward function

Answer 3

An ML method based on neural networks with multiple layers

Answer 4

A type of ML that learns an optimal policy to maximize the reward function without needing a labeled dataset. ## Footnote Reinforcement learning differs from supervised and unsupervised learning by focusing on reward signals rather than data relationships.

Answer 5

An ML method based on neural networks with several layers, known for its flexibility and ability to learn from errors. ## Footnote Deep learning can be applied to supervised, unsupervised, or reinforcement learning tasks.

Answer 6

Models with a finite number of parameters, assuming the underlying probability distribution is defined by the parameter set. ## Footnote Non-parametric models do not have a predefined structure and generally require more data.

Answer 7

* Customer service and support * Business support in consultancy/portfolio management * Compliance * Risk Management * Fraud detection * Operational efficiency ## Footnote These uses illustrate AI's potential impact across various functions.

Answer 8

AI is primarily used for specific tasks rather than representing a revolutionary transformation in investment practices. ## Footnote Few funds have developed a fully end-to-end AI-based investment process.

Answer 9

Relatively simple algorithms using limited client information, such as investment horizon and risk tolerance. ## Footnote The integration of advanced AI models into robo-advisors remains uncertain.

Answer 10

True ## Footnote AI helps reduce market impact of large orders and minimize settlement failures.

Answer 11

Asset price signals to identify investment opportunities. ## Footnote High-frequency traders and quantitative hedge funds commonly use these algorithms.

Answer 12

securities lending ## Footnote AI also predicts which securities may become hard-to-borrow.

Answer 13

To predict the probability that a transaction will not be settled and to optimally allocate resources. ## Footnote Central securities depositories and brokers utilize ML for this purpose.

Answer 14

* Explainability * Interconnection and systemic risk * Lack of accountability and transparency * Algorithmic bias * Operational risk and data privacy * Data quality and model risk ## Footnote These risks necessitate ongoing monitoring and understanding.

Answer 15

Firms must act in the best interests of clients, ensure transparency, and provide adequate governance and risk management. ## Footnote This includes training staff on AI risks and ensuring data quality.

Answer 16

Natural Language Processing is used to process large amounts of text to identify unstructured information and conduct market sentiment analysis. ## Footnote NLP is becoming prevalent across all sectors in capital markets.

Answer 17

False ## Footnote The current regulatory framework does not present significant barriers to AI deployment.

Answer 18

* Complexity * Explainability * Lack of transparency ## Footnote These concerns are crucial for effective supervision of AI technologies.

Fintech (INstitutions) Flashcards

(42 cards)