An official website of the United States government
As practitioners increasingly deploy machine learning models in critical domains such as health care, finance, and policy, it becomes vital to ensure that domain experts function effectively alongside these models. Explainability is one way to bridge the gap between human decision-makers and machine learning models. However, most of the existing work on explainability focuses on one-off, static explanations like feature importances or rule lists. These sorts of explanations may not be sufficient for many use cases that require dynamic, continuous discovery from stakeholders. In the literature, few works ask decision-makers about the utility of existing explanations and other desiderata they would like to see in an explanation going forward. In this work, we address this gap and carry out a study where we interview doctors, healthcare professionals, and policymakers about their needs and desires for explanations. Our study indicates that decision-makers would strongly prefer interactive explanations in the form of natural language dialogues. Domain experts wish to treat machine learning models as "another colleague", i.e., one who can be held accountable by asking why they made a particular decision through expressive and accessible natural language interactions. Considering these needs, we outline a set of five principles researchers should follow when designing interactive explanations as a starting place for future work. Further, we show why natural language dialogues satisfy these principles and are a desirable way to build interactive explanations. Next, we provide a design of a dialogue system for explainability and discuss the risks, trade-offs, and research opportunities of building these systems. Overall, we hope our work serves as a starting place for researchers and engineers to design interactive explainability systems.
more »
« less
Towards Interactive Curation & Automatic Tuning of ML Pipelines
Democratizing Data Science requires a fundamental rethinking of the way data analytics and model discovery is done. Available tools for analyzing massive data sets and curating machine learning models are limited in a number of fundamental ways. First, existing tools require well-trained data scientists to select the appropriate techniques to build models and to evaluate their outcomes. Second, existing tools require heavy data preparation steps and are often too slow to give interactive feedback to domain experts in the model building process, severely limiting the possible interactions. Third, current tools do not provide adequate analysis of statistical risk factors in the model development. In this work, we present the first iteration of QuIC-M (pronounced quick-m), an interactive human-in-the-loop data exploration and model building suite. The goal is to enable domain experts to build the machine learning pipelines an order of magnitude faster than machine learning experts while having model qualities comparable to expert solutions.
more »
« less
- Award ID(s):
- 1740741
- PAR ID:
- 10066177
- Date Published:
- Journal Name:
- Proceedings of the Second Workshop on Data Management for End-to-End Machine Learning
- Page Range / eLocation ID:
- 1 to 4
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Statistical knowledge and domain expertise are key to extract actionable insights out of data, yet such skills rarely coexist together. In Machine Learning, high-quality results are only attainable via mindful data preprocessing, hyperparameter tuning and model selection. Domain experts are often overwhelmed by such complexity, de-facto inhibiting a wider adoption of ML techniques in other fields. Existing libraries that claim to solve this problem, still require well-trained practitioners. Those frameworks involve heavy data preparation steps and are often too slow for interactive feedback from the user, severely limiting the scope of such systems. In this paper we present Alpine Meadow, a first Interactive Automated Machine Learning tool. What makes our system unique is not only the focus on interactivity, but also the combined systemic and algorithmic design approach; on one hand we leverage ideas from query optimization, on the other we devise novel selection and pruning strategies combining cost-based Multi-Armed Bandits and Bayesian Optimization. We evaluate our system on over 300 datasets and compare against other AutoML tools, including the current NIPS winner, as well as expert solutions. Not only is Alpine Meadow able to significantly outperform the other AutoML systems while --- in contrast to the other systems --- providing interactive latencies, but also outperforms in 80% of the cases expert solutions over data sets we have never seen before.more » « less
-
Building a skilled cybersecurity workforce is paramount to building a safer digital world. However, the diverse skill set, constantly emerging vulnerabilities, and deployment of new cyber threats make learning cybersecurity challenging. Traditional education methods struggle to cope with cybersecurity's rapidly evolving landscape and keep students engaged and motivated. Different studies on students' behaviors show that an interactive mode of education by engaging through a question-answering system or dialoguing is one of the most effective learning methodologies. There is a strong need to create advanced AI-enabled education tools to promote interactive learning in cybersecurity. Unfortunately, there are no publicly available standard question-answer datasets to build such systems for students and novice learners to learn cybersecurity concepts, tools, and techniques. The education course material and online question banks are unstructured and need to be validated and updated by domain experts, which is tedious when done manually. In this paper, we propose CyberGen, a novel unification of large language models (LLMs) and knowledge graphs (KG) to generate the questions and answers for cybersecurity automatically. Augmenting the structured knowledge from knowledge graphs in prompts improves factual reasoning and reduces hallucinations in LLMs. We used the knowledge triples from cybersecurity knowledge graphs (AISecKG) to design prompts for ChatGPT and generate questions and answers using different prompting techniques. Our question-answer dataset, CyberQ, contains around 4k pairs of questions and answers. The domain expert manually evaluated the random samples for consistency and correctness. We train the generative model using the CyberQ dataset for question answering task.more » « less
-
Many analytic tools have been developed to discover knowledge from student data. However, the knowledge discovery process requires advanced analytical modelling skills, making it the province of data scientists. This impedes the ability of educational leaders, professors, and advisors to engage with the knowledge discovery process directly. As a result, it is challenging for analysis to take advantage of domain expertise, making its outcome often neither interesting nor useful. Usually the outcome produced from such analytic tools is static, preventing domain experts from exploring different hypotheses by changing data models or predictive models inside the tool. We have developed a framework for interactive and exploratory learning analytics which begins to address these challenges. We engaged in data exploration and hypotheses generation with our university domain experts by conducting two focus groups. We used the findings of these focus groups to validate our framework, arguing that it enables domain experts to explore the data, analysis and interpretation of student data to discover useful and interesting knowledge.more » « less
-
Build systems are responsible for building software correctly and quickly. Unfortunately, traditional build tools like make are correct and fast only when developers precisely enumerate dependencies for every incremental build step. Forward build systems improve correctness over traditional build tools by discovering dependencies automatically, but existing forward build tools have two fundamental flaws. First, they are incorrect; existing forward build tools miss dependencies because their models of system state are incomplete. Second, they rely on users to manually specify incremental build steps, increasing the programmer burden for fast builds. This paper introduces Riker, a forward build system that guarantees fast, correct builds. Riker builds are easy to specify; in many cases a single command such as gcc *.c suffices. From these simple specifications, Riker automatically discovers fast incremental rebuild opportunities. Riker models the entire POSIX filesystem—not just files, but directories, pipes, and so on. This model guarantees that every dependency is checked on every build so every output is correct. We use Riker to build 14 open source packages including LLVM and memcached. Riker incurs a median overhead of 8.8% on the initial full build. On average, Riker’s incremental builds realize 94% of make’s incremental speedup with no manual effort and no risk of errors.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.1145/3209889.3209891
