Amazon AI researchers present MEMENTO: a neural model to estimate the effects of an individual treatment for multiple treatments

0

Estimating the effect of a binary intervention from historical observational data has been extensively studied in the causal inference literature. However, the main objective of these surveys has been to estimate the typical impact of the intervention. It is possible that the variation in response to the intervention (treatment) between individuals and groups within a population is lost in the calculation of the mean treatment effect (ATE). Therefore, it is crucial to calculate the impact of treatment individually.

Most causal estimation methods have focused on a situation where treatments are turned on or off. Making predictions about causal effects when there are many mutually incompatible treatments is, by extension, a significant challenge in many fields.

Without having direct access to the process that gave rise to the treatments, observational data includes past interventions, their effects, and perhaps additional context. For cases with more than two distinct values ​​in a discrete and finite set of treatments, prediction is particularly relevant. Confounding is an important element in inferring causal effects from observational data. A confounder of the effect of treatment on its outcome is a variable that influences therapy and its outcome. Controlling for such a confounder is conventional practice if its impact can be quantified.

Most of the work on the estimation of the impact of treatments is devoted to binary treatments, which does not easily generalize to the case of multiple treatments. A new study from Amazon provides a method for estimating the individual-level effects of treatments when those effects are discontinuous and finite.

The researchers provide expanded definitions of factual and counterfactual prediction errors in the presence of many treatments, extending this method to multi-treatment scenarios and determining an upper constraint on the total amount of factual and counterfactual losses.

By minimizing an upper bound on the total of factual and counterfactual losses, the method obtains consistent representations of confounding factors between treatment types.

To minimize this loss, the researchers suggest using the neural model. Since the proposed technique and system memorize processes (or events) through the construction of representations, they refer to it as MEMENTO. As a framework, MEMENTO provides both a loss function and a model that seeks to minimize this function. Therefore, any underlying modeling technique that can optimize the loss can be used.

Amazon uses MEMENTO to determine the MOQ or minimum order quantity of a product. The team compares the results of competing algorithms on public and mock datasets to ensure reproducibility. In some use cases, experiments on real, semi-synthetic datasets reveal that MEMENTO can outperform established approaches for multi-processing situations by approximately 10%.

With the implementation of MEMENTO on the MOQ issue, the system went into full-scale production on March 21. According to an A/B experiment conducted in a new market, the results show that MEMENTO has a 4.7% impact on reducing shipping costs. when applied to the question of MOQ.

This methodology could be used in various contexts inside and outside of Amazon. In the future, the team hopes to expand their study to include continuous treatments and find ways to provide accurate estimates when key variables are missing.

This Article is written as a research summary article by Marktechpost Staff based on the research paper 'MEMENTO: Neural model for estimating individual treatment effects for multiple treatments'. All Credit For This Research Goes To Researchers on This Project. Check out the paper.

Please Don't Forget To Join Our ML Subreddit


Tanushree Shenwai is an intern consultant at MarktechPost. She is currently pursuing her B.Tech from Indian Institute of Technology (IIT), Bhubaneswar. She is a data science enthusiast and has a keen interest in the scope of application of artificial intelligence in various fields. She is passionate about exploring new technological advancements and applying them to real life.


Share.

About Author

Comments are closed.