Tony Sun

Tony Y. Sun, Shreyas Bhave, Jaan Altosaar et al.

Disease identification is a core, routine activity in observational health research. Cohorts impact downstream analyses, such as how a condition is characterized, how patient risk is defined, and what treatments are studied. It is thus critical to ensure that selected cohorts are representative of all patients, independently of their demographics or social determinants of health. While there are multiple potential sources of bias when constructing phenotype definitions which may affect their fairness, it is not standard in the field of phenotyping to consider the impact of different definitions across subgroups of patients. In this paper, we propose a set of best practices to assess the fairness of phenotype definitions. We leverage established fairness metrics commonly used in predictive models and relate them to commonly used epidemiological cohort description metrics. We describe an empirical study for Crohn's disease and diabetes type 2, each with multiple phenotype definitions taken from the literature across two sets of patient subgroups (gender and race). We show that the different phenotype definitions exhibit widely varying and disparate performance according to the different fairness metrics and subgroups. We hope that the proposed best practices can help in constructing fair and inclusive phenotype definitions.

Ian Covert, Tony Sun, James Zou et al.

Vision language models (VLMs) have seen growing adoption in recent years, but many still struggle with basic spatial reasoning errors. We hypothesize that this is due to VLMs adopting pre-trained vision backbones, specifically vision transformers (ViTs) trained with image-level supervision and minimal inductive biases. Such models may fail to encode the class contents at each position in the image, and our goal is to resolve this with a vision backbone that effectively captures both local and global image semantics. Our main insight is that we do not require new supervision to learn this capability - pre-trained models contain significant knowledge of local semantics that we can extract and use for scalable self-supervision. We propose a new efficient post-training stage for ViTs called locality alignment and a novel fine-tuning procedure called MaskEmbed that uses a masked reconstruction loss to learn semantic contributions for each image patch. We first evaluate locality alignment with a vision-only benchmark, finding that it improves a model's performance at patch-level semantic segmentation, especially for strong backbones trained with image-caption pairs (e.g., CLIP and SigLIP). We then train a series of VLMs with and without locality alignment, and show that locality-aligned backbones improve performance across a range of benchmarks, particularly ones that involve spatial understanding (e.g., RefCOCO, OCID-Ref, TallyQA, VSR, AI2D). Overall, we demonstrate that we can efficiently learn local semantic extraction via a locality alignment stage, and that this procedure benefits VLM training recipes that use off-the-shelf vision backbones.

Ethan Li, Anders Boesen Lindbo Larsen, Chen Zhang et al. · apple-ml, cmu

We introduce two multilingual, multimodal foundation language models that power Apple Intelligence features across Apple devices and services: i a 3B-parameter on-device model optimized for Apple silicon through architectural innovations such as KV-cache sharing and 2-bit quantization-aware training; and ii a scalable server model built on a novel Parallel-Track Mixture-of-Experts PT-MoE transformer that combines track parallelism, mixture-of-experts sparse computation, and interleaved global-local attention to deliver high quality with competitive cost on Apple's Private Cloud Compute platform. Both models are trained on large-scale multilingual and multimodal datasets sourced via responsible web crawling, licensed corpora, and high-quality synthetic data, then further refined with supervised fine-tuning and reinforcement learning on a new asynchronous platform. The resulting models support several additional languages while understanding images and executing tool calls. In public benchmarks and human evaluations, both the server model and the on-device model match or surpass comparably sized open baselines. A new Swift-centric Foundation Models framework exposes guided generation, constrained tool calling, and LoRA adapter fine-tuning, allowing developers to integrate these capabilities with a few lines of code. The latest advancements in Apple Intelligence models are grounded in our Responsible AI approach with safeguards like content filtering and locale-specific evaluation, as well as our commitment to protecting our users' privacy with innovations like Private Cloud Compute.

Tony Sun, Kellie Webster, Apu Shah et al.

Responsible development of technology involves applications being inclusive of the diverse set of users they hope to support. An important part of this is understanding the many ways to refer to a person and being able to fluently change between the different forms as needed. We perform a case study on the singular they, a common way to promote gender inclusion in English. We define a re-writing task, create an evaluation benchmark, and show how a model can be trained to produce gender-neutral English with <1% word error rate with no human-labeled data. We discuss the practical applications and ethical considerations of the task, providing direction for future work into inclusive natural language systems.

Tony Y. Sun, Oliver J. Bear Don't Walk, Jennifer L. Chen et al.

Sex and gender-based healthcare disparities contribute to differences in health outcomes. We focus on time to diagnosis (TTD) by conducting two large-scale, complementary analyses among men and women across 29 phenotypes and 195K patients. We first find that women are consistently more likely to experience a longer TTD than men, even when presenting with the same conditions. We further explore how TTD disparities affect diagnostic performance between genders, both across and persistent to time, by evaluating gender-agnostic disease classifiers across increasing diagnostic information. In both fairness analyses, the diagnostic process favors men over women, contradicting the previous observation that women may demonstrate relevant symptoms earlier than men. These analyses suggest that TTD is an important yet complex aspect when studying gender disparities, and warrants further investigation.

Andrew Gaut, Tony Sun, Shirlyn Tang et al.

Recent developments in Neural Relation Extraction (NRE) have made significant strides towards Automated Knowledge Base Construction (AKBC). While much attention has been dedicated towards improvements in accuracy, there have been no attempts in the literature to our knowledge to evaluate social biases in NRE systems. We create WikiGenderBias, a distantly supervised dataset with a human annotated test set. WikiGenderBias has sentences specifically curated to analyze gender bias in relation extraction systems. We use WikiGenderBias to evaluate systems for bias and find that NRE systems exhibit gender biased predictions and lay groundwork for future evaluation of bias in NRE. We also analyze how name anonymization, hard debiasing for word embeddings, and counterfactual data augmentation affect gender bias in predictions and performance.

Tony Sun, Andrew Gaut, Shirlyn Tang et al.

As Natural Language Processing (NLP) and Machine Learning (ML) tools rise in popularity, it becomes increasingly vital to recognize the role they play in shaping societal biases and stereotypes. Although NLP models have shown success in modeling various applications, they propagate and may even amplify gender bias found in text corpora. While the study of bias in artificial intelligence is not new, methods to mitigate gender bias in NLP are relatively nascent. In this paper, we review contemporary studies on recognizing and mitigating gender bias in NLP. We discuss gender bias based on four forms of representation bias and analyze methods recognizing gender bias. Furthermore, we discuss the advantages and drawbacks of existing gender debiasing methods. Finally, we discuss future studies for recognizing and mitigating gender bias in NLP.