Campus Event Calendar
Event Entry
What and Who
On Fairness, Invariance and Memorization in Machine Decision and DeepLearning Algorithms
Till Speicher
Max Planck Institute for Software Systems
SWS Student Defense Talks - Thesis Proposal
AG 1, AG 2, AG 3, INET, AG 4, AG 5, D6, SWS, RG1, MMCI
AG Audience
English
Note: We use this to send email in the morning.
Date, Time and Location
Tuesday, 26 March 2024
12:00
60 Minutes
E1 5
005
Saarbrücken
Abstract
As learning algorithms become more and more capable, they are used to tackle an increasingly large spectrum of tasks. Their applications range from understanding images, speech and natural language to making socially impactful decisions, such as about people's eligibility for loans and jobs. Therefore, it is important to better understand both the consequences of algorithmic decisions as well as the mechanisms by which algorithms arrive at their outputs. Of particular interest in
this regard are fairness when algorithmic decisions impact peoples lives, as well as the behavior of deep learning algorithms, the most powerful but also opaque type of learning algorithm.
To this end, this thesis makes two contributions: First, we study fairness in algorithmic decision making. At a conceptual level, we introduce a metric for measuring unfairness in algorithmic decisions based on inequality indices from the economics literature. We show that this metric can be used to decompose the overall unfairness for a given set of users into between- and within-subgroup components and highlight potential tradeoffs between them as well as between fairness and accuracy. At an empirical level, we demonstrate the necessity for
studying fairness in algorithmically controlled systems by exposing the potential for discrimination that is enabled by Facebook's advertising platform. In this context, we demonstrate how advertisers can target ads to exclude users belonging to protected sensitive groups, a practice that is illegal in domains such as housing, employment and finance, and highlight the necessity for better mitigation methods.
The second contribution of this thesis is aimed at better understanding the mechanisms governing the behavior of deep learning algorithms. First, we study the role that invariance plays in learning useful representations. We show that the set of invariances possessed by representations is of critical importance in determining whether they are useful for downstream tasks, more important than many other factors commonly considered to determine transfer performance. Second, we investigate memorisation in large language models, which have recently become very popular. By training models to memorise random strings, we surface a rich and surprising set of dynamics during the memorisation process. We find that models undergo two phases during memorisation, that strings with lower entropy are harder to memorise, that the memorisation dynamics evolve during repeated memorisation and that models can recall tokens in random strings with only a very restricted amount of information.
this regard are fairness when algorithmic decisions impact peoples lives, as well as the behavior of deep learning algorithms, the most powerful but also opaque type of learning algorithm.
To this end, this thesis makes two contributions: First, we study fairness in algorithmic decision making. At a conceptual level, we introduce a metric for measuring unfairness in algorithmic decisions based on inequality indices from the economics literature. We show that this metric can be used to decompose the overall unfairness for a given set of users into between- and within-subgroup components and highlight potential tradeoffs between them as well as between fairness and accuracy. At an empirical level, we demonstrate the necessity for
studying fairness in algorithmically controlled systems by exposing the potential for discrimination that is enabled by Facebook's advertising platform. In this context, we demonstrate how advertisers can target ads to exclude users belonging to protected sensitive groups, a practice that is illegal in domains such as housing, employment and finance, and highlight the necessity for better mitigation methods.
The second contribution of this thesis is aimed at better understanding the mechanisms governing the behavior of deep learning algorithms. First, we study the role that invariance plays in learning useful representations. We show that the set of invariances possessed by representations is of critical importance in determining whether they are useful for downstream tasks, more important than many other factors commonly considered to determine transfer performance. Second, we investigate memorisation in large language models, which have recently become very popular. By training models to memorise random strings, we surface a rich and surprising set of dynamics during the memorisation process. We find that models undergo two phases during memorisation, that strings with lower entropy are harder to memorise, that the memorisation dynamics evolve during repeated memorisation and that models can recall tokens in random strings with only a very restricted amount of information.
Contact
Susanne Girard
+49 631 9303 9605
--email hidden
passcode not visible
logged in users only
Susanne Girard, 03/25/2024 14:06 -- Created document.