A5068178240- Artificial Intelligence in Law
- Natural Language Processing Techniques
- Topic Modeling
- Web Data Mining and Analysis
- Legal Education and Practice Innovations
- Semantic Web and Ontologies
- Information Retrieval and Search Behavior
- Privacy-Preserving Technologies in Data
- Artificial Intelligence in Healthcare and Education
- Domain Adaptation and Few-Shot Learning
- Adversarial Robustness in Machine Learning
- Advanced Neural Network Applications
- Spam and Phishing Detection
- Comparative and International Law Studies
- Legal Language and Interpretation
- Law, AI, and Intellectual Property
- COVID-19 Digital Contact Tracing
- Machine Learning and Data Classification
- Health disparities and outcomes
- COVID-19 epidemiological studies
- Stochastic Gradient Optimization Techniques
- Privacy, Security, and Data Protection
- Internet Traffic Analysis and Secure E-voting
- Web Application Security Vulnerabilities
- American Constitutional Law and Politics
Stanford University
2015-2025
Stanford Medicine
2023
Carnegie Mellon University
2018
Palo Alto University
2013-2014
AI is undergoing a paradigm shift with the rise of models (e.g., BERT, DALL-E, GPT-3) that are trained on broad data at scale and adaptable to wide range downstream tasks. We call these foundation underscore their critically central yet incomplete character. This report provides thorough account opportunities risks models, ranging from capabilities language, vision, robotics, reasoning, human interaction) technical principles(e.g., model architectures, training procedures, data, systems,...
This JAMA Forum discusses the possibilities, limitations, and risks of physician use large language models (such as ChatGPT) along with improvements required to improve accuracy technology.
The advent of large language models (LLMs) and their adoption by the legal community has given rise to question: what types reasoning can LLMs perform? To enable greater study this question, we present LegalBench: a collaboratively constructed benchmark consisting 162 tasks covering six different reasoning. LegalBench was built through an interdisciplinary process, in which collected designed hand-crafted professionals. Because these subject matter experts took leading role construction,...
The authors review challenges arising in malpractice litigation related to software errors inform health care organizations and physicians about liability risk from AI adoption strategies mitigate risk.
We present one-shot federated learning, where a central server learns global model over network of devices in single round communication. Our approach - drawing on ensemble learning and knowledge aggregation achieves an average relative gain 51.5% AUC local baselines comes within 90.1% the (unattainable) ideal. discuss these methods identify several promising directions future work.
While self-supervised learning has made rapid advances in natural language processing, it remains unclear when researchers should engage resource-intensive domain-specific pretraining (domain pretraining). The law, puzzlingly, yielded few documented instances of substantial gains to domain spite the fact that legal is widely seen be unique. We hypothesize these existing results stem from NLP tasks are too easy and fail meet conditions for can help. To address this, we first present CaseHOLD...
Language models (LMs) are becoming the foundation for almost all major language technologies, but their capabilities, limitations, and risks not well understood. We present Holistic Evaluation of Models (HELM) to improve transparency models. First, we taxonomize vast space potential scenarios (i.e. use cases) metrics desiderata) that interest LMs. Then select a broad subset based on coverage feasibility, noting what's missing or underrepresented (e.g. question answering neglected English...
Large language models (LLMs) transfer well to new tasks out-of-the-box simply given a natural prompt that demonstrates how perform the task and no additional training. Prompting is brittle process wherein small modifications can cause large variations in model predictions, therefore significant effort dedicated towards designing painstakingly "perfect prompt" for task. To mitigate high degree of involved prompt-design, we instead ask whether producing multiple effective, yet imperfect,...
We explore machine learning methods for AC Optimal Powerflow (ACOPF) - the task of optimizing power generation in a transmission network according while respecting physical and engineering constraints. present two formulations ACOPF as problem: 1) an end-to-end prediction where we directly predict optimal generator settings, 2) constraint set active constraints solution. validate these approaches on benchmark grids.
Anonymized smartphone-based mobility data has been widely adopted in devising and evaluating COVID-19 response strategies such as the targeting of public health resources. Yet little attention paid to measurement validity demographic bias, due part lack documentation about which users are represented well challenge obtaining ground truth on unique visits demographics. We illustrate how linking large-scale administrative can enable auditing for bias absence information labels. More precisely,...
One concern with the rise of large language models lies their potential for significant harm, particularly from pretraining on biased, obscene, copyrighted, and private information. Emerging ethical approaches have attempted to filter material, but such been ad hoc failed take context into account. We offer an approach filtering grounded in law, which has directly addressed tradeoffs material. First, we gather make available Pile Law, a 256GB (and growing) dataset open-source...
A challenge for named entity disambiguation (NED), the task of mapping textual mentions to entities in a knowledge base, is how disambiguate that appear rarely training data, termed tail entities. Humans use subtle reasoning patterns based on facts, relations, and types unfamiliar Inspired by these patterns, we introduce Bootleg, self-supervised NED system explicitly grounded disambiguation. We define core disambiguation, create learning procedure encourage model learn show weak supervision...
Large language models (LLMs) are increasingly used in applications where LLM inputs may span many different tasks. Recent work has found that the choice of is consequential, and LLMs be good for input samples. Prior approaches have thus explored how engineers might select an to use each sample (i.e. _routing_). While existing routing methods mostly require training auxiliary on human-annotated data, our explores whether it possible perform _unsupervised_ routing. We propose SMOOTHIE, a weak...
Weak supervision (WS) is a popular approach for label-efficient learning, leveraging diverse sources of noisy but inexpensive weak labels to automatically annotate training data. Despite its wide usage, WS and practical value are challenging benchmark due the many knobs in setup, including: data sources, labeling functions (LFs), aggregation techniques (called label models), end model pipelines. Existing evaluation suites tend be limited, focusing on particular components or specialized use...
Can foundation models be guided to execute tasks involving legal reasoning? We believe that building a benchmark answer this question will require sustained collaborative efforts between the computer science and communities. To end, short paper serves three purposes. First, we describe how IRAC-a framework scholars use distinguish different types of reasoning-can guide construction Foundation Model oriented benchmark. Second, present seed set 44 built according framework. discuss initial...
While self-supervised learning has made rapid advances in natural language processing, it remains unclear when researchers should engage resource-intensive domain-specific pretraining (domain pretraining). The law, puzzlingly, yielded few documented instances of substantial gains to domain spite the fact that legal is widely seen be unique. We hypothesize these existing results stem from NLP tasks are too easy and fail meet conditions for can help. To address this, we first present CaseHOLD...
Retrieval pipelines-an integral component of many machine learning systems-perform poorly in domains where documents are long (e.g., 10K tokens or more) and identifying the relevant document requires synthesizing information across entire text. Developing long-context retrieval encoders suitable for these raises three challenges: (1) how to evaluate performance, (2) pretrain a base language model represent both short contexts (corresponding queries) documents), (3) fine-tune this under batch...
Recent work has shown that language models' (LMs) prompt-based learning capabilities make them well suited for automating data labeling in domains where manual annotation is expensive. The challenge while writing an initial prompt cheap, improving a costly -- practitioners often require significant labeled order to evaluate the impact of modifications. Our asks whether it possible improve without additional data. We approach this problem by attempting modify predictions prompt, rather than...
The advent of large language models (LLMs) and their adoption by the legal community has given rise to question: what types reasoning can LLMs perform? To enable greater study this question, we present LegalBench: a collaboratively constructed benchmark consisting 162 tasks covering six different reasoning. LegalBench was built through an interdisciplinary process, in which collected designed hand-crafted professionals. Because these subject matter experts took leading role construction,...
Scholarship on U.S. litigation and civil procedure has scarcely studied the role of private enforcement in states. Over past two decades, scholars have established that, almost uniquely world, often relies parties rather than administrative agencies to enforce important statutory provisions. Take your pick any area American governance you will find rights action: environmental law, rights, employment discrimination, antitrust, consumer protection, business competition, securities fraud, so...
short-paper Share on Course specific search engines: a study in incorporating context into Author: Neel Guha Henry M. Gunn High School, Palo Alto, CA, USA USAView Profile Authors Info & Claims ESAIR '13: Proceedings of the sixth international workshop Exploiting semantic annotations information retrievalOctober 2013 Pages 33–36https://doi.org/10.1145/2513204.2513216Published:28 October 2013Publication History 0citation84DownloadsMetricsTotal Citations0Total Downloads84Last 12 Months3Last 6...