Motivation

Machine Learning Systems

Machine Learning Systems

1. Preference Dynamics
2. Implications: Personalization & Harm
   
3. Retention Dynamics

Outline

How do we design reliable algorithms that account for user dynamics?

Setting: Preference Dynamics

Interests may be impacted by recommended content

Setting: Preference Dynamics

$a_t$

$y_t = \langle s_t, a_t\rangle + w_t$

Interests may be impacted by recommended content

underlies factorization-based methods

Setting: Preference Dynamics

$a_t$

$y_t = \langle s_t, a_t\rangle + w_t$

Interests may be impacted by recommended content

underlies factorization-based methods

Preference Dynamics

items $a_t\in\mathcal A\subseteq \mathcal S^{d-1}$

$y_t = \langle s_t, a_t\rangle + w_t$

$s_{t+1} = f_t(s_t, a_t)$

preferences $s\in\mathcal S^{d-1}$

Preference Dynamics

items $a_t\in\mathcal A\subseteq \mathcal S^{d-1}$

$y_t = \langle s_t, a_t\rangle + w_t$

$s_{t+1} \propto s_t + \eta_t a_t$

preferences $s\in\mathcal S^{d-1}$

Assimilation: interests may become more similar to recommended content

Preference Dynamics

items $a_t\in\mathcal A\subseteq \mathcal S^{d-1}$

$y_t = \langle s_t, a_t\rangle + w_t$

Biased Assimilation: interest update is proportional to affinity

$s_{t+1} \propto s_t + \eta_t\langle s_t, a_t\rangle a_t$

preferences $s\in\mathcal S^{d-1}$

Prior Work

2. Biased assimilation

$s_{t+1} \propto s_t + \eta_t\langle s_t, a_t\rangle a_t$

When recommendations are made globally, the outcomes differ:

1. Assimilation

$s_{t+1} \propto s_t + \eta_t a_t$

Personalized Recommendations

Regardless of whether assimilation is biased,

$s_{t+1} \propto s_t + \eta_t\langle s_t, a_t\rangle a_t$

$s_{t+1} \propto s_t + \eta_t a_t$

Personalized Recommendations

Regardless of whether assimilation is biased,

$s_{t+1} \propto s_t + \eta_t\langle s_t, a_t\rangle a_t$

$s_{t+1} \propto s_t + \eta_t a_t$

Personalized Recommendations

Regardless of whether assimilation is biased,

$s_{t+1} \propto s_t + \eta_t\langle s_t, a_t\rangle a_t$

$s_{t+1} \propto s_t + \eta_t a_t$

Personalized Recommendations

Regardless of whether assimilation is biased,

$s_{t+1} \propto s_t + \eta_t\langle s_t, a_t\rangle a_t$

$s_{t+1} \propto s_t + \eta_t a_t$

Harmful Recommendations

Preference dynamics lead to a new perspective on harm

𝅗𝅥

Harmful Recommendations

Due to preference dynamics, there may be downstream harm, even when no harmful content is recommended

♪

♫

𝅘𝅥

𝅗𝅥

𝅘𝅥𝅯

♫

𝅗𝅥

Recommendation: ♫

Harmful Recommendations

Due to preference dynamics, there may be downstream harm, even when no harmful content is recommended

♪

♫

𝅘𝅥

𝅗𝅥

𝅘𝅥𝅯

♫

𝅗𝅥

Recommendation: ♫

Recommendation: 𝅘𝅥𝅯

Harmful Recommendations

Due to preference dynamics, there may be downstream harm, even when no harmful content is recommended

Recommendation: ♫

Recommendation: 𝅘𝅥𝅯

This motivates a new recommendation objective which takes into account the probability of future harm [CDEIKW23]

User Choice & Retention

Even if individual preferences are immutable, population level effects may be observed due to retention dynamics

♫

𝅘𝅥

♫

𝅗𝅥

User Choice & Retention

Even if individual preferences are immutable, population level effects may be observed due to retention dynamics

𝅘𝅥

The dynamic of retention & specialization can lead to representation disparity (Hashimoto et al.) and segmentation [DCRMF23]

Example: linear regression with users      and         

User Choice & Retention

Even if individual preferences are immutable, population level effects may be observed due to retention dynamics

𝅘𝅥

The dynamic of retention & specialization can lead to representation disparity (Hashimoto et al.) and segmentation [DCRMF23]

Example: linear regression with users      and         

User Choice & Retention

Even if individual preferences are immutable, population level effects may be observed due to retention dynamics

The dynamic of retention & specialization can lead to representation disparity (Hashimoto et al.) and segmentation [DCRMF23]

Example: linear regression with users      and         

• services 1 and 2

Conclusion & Discussion

Other References

• Gaitonde, Kleinberg, Tardos, 2021. Polarization in geometric opinion dynamics. EC.
• Hązła, Jin, Mossel, Ramnarayan, 2019. A geometric model of opinion polarization. arXiv:1910.05274.

• Hashimoto, Srivastava, Namkoong, Liang, 2018. Fairness Without Demographics in Repeated Loss Minimization. ICML.

References