Recent Developments in Research on Fairness 2019
2019-07-08..2019-07-10
Berkeley, United States
| title | author | length |
|---|---|---|
| What is Your Data Worth? Equitable Data Valuation in Machine Learning | James Zou | 44m |
Defines data Shapley value as the marginal contribution of each training point averaged over all subsets and shows it outperforms the common leave-one-out… | ||
| What It Takes to Control Societal Bias in Natural Language Processing | Kai-Wei Chang | 50m |
Kai-Wei Chang (UCLA)
https://simons.berkeley.edu/talks/what-it-takes-control-societal-bias-natural-language-processing
Recent Developments in Research on… | ||
| Flexibly Fair Representation Learning by Disentanglement | Rich Zemel | 37m |
Proposes a single encoder that outputs a Z and B latent split (Z for non‑sensitive info, B for per‑attribute latents), trained with VAE reconstruction plus… | ||
| Bridging AI and HCI: Incorporating Human Values into the Development of AI Technologies | Haiyi Zhu | 48m |
A Wikipedia recruitment system is built and field-deployed to match newcomers to wiki projects using four models (two interest-based — rule and category… | ||
| Procedural Justice in Algorithmic Fairness: Transparency, Control and Participation Perspectives | Min Kyung Lee | 44m |
Argues that procedural fairness—transparency, control and participation—matters as much as outcome metrics; compares a fair-division algorithm (used via Split… | ||
| A Multi-Group Perspective on Fairness | Guy N. Rothblum | 47m |
Claims that multi calibration guarantees exist for any class C and that small, efficient predictors can be learned from labeled outcomes; algorithm starts with… | ||
| Evidence-Based Rankings | Michael Kim | 45m |
Defines domination compatibility and a stronger evidence consistency criterion for rankings, proves that a ranking induced by a C-multi-accurate predictor… | ||
| The Measure and Mismeasure of Fairness | Sharad C. Goel | 46m |
Argues that common algorithmic risk scores (logistic regressions, threshold rules, confusion-matrix metrics) can mislead: group-based formal measures like… | ||
| Disentangling Influence: Using Disentangled Representations to Audit Model Predictions | Charlie Marx | 27m |
Method reduces indirect influence auditing to direct influence by learning a disentangled latent where protected factor is independent, then composing its… | ||
| Representation via Representations: Improving Generalization to Unseen Populations | Frances Ding | 38m |
Proposes training an encoder+predictor+decoder with an adversary that forces representations to remove which-study signals so predictors rely on invariant… | ||
| The Social Cost of Strategic Classification | John Miller | 36m |
Argues that raising classifier thresholds to optimize institutional accuracy via the Stackelberg equilibrium forces higher expected costs on truly qualified… | ||
| Eliciting and Enforcing Subjective Individual Fairness | Aaron Roth | 46m |
Proposes an algorithm that enforces pairwise subjective similarity constraints without assuming any metric, trading off classification error against the… | ||
| Individual Fairness in Hindsight | Swati Gupta | 31m |
Defines individual fairness in hindsight as a Lipschitz-style rule that only imposes past-based lower bounds on current decisions so decisions can… | ||
| (Improving) Muddled Information | Alex Frankel | 46m |
Heterogeneous gaming ability causes manipulable signals to decouple from true types, so raising incentives reduces information: a two-by-two model and a… | ||
| Equal Opportunity in Online Classification with Partial Feedback | Yahav Bechavod | 34m |
Online classification with one-sided feedback (lend/deny), cast to contextual bandits by adding a constant loss for denied cases so the reduction is… | ||
| Journalist-in-Residence Talk: The Limits of Fairness | Konstantin Kakaes | 01h 10m |
Argues that fairness metrics and better models often miss the real harms: algorithms reshape norms (example: Gmail autocomplete nudges acceptable replies)… | ||
| nothing matches | ||