Related papers: PRE-Share Data: Assistance Tool for Resource-aware…
To train modern large DNN models, pipeline parallelism has recently emerged, which distributes the model across GPUs and enables different devices to process different microbatches in pipeline. Earlier pipeline designs allow multiple…
The process of data analysis, especially in GUI-based analytics systems, is highly exploratory. The user iteratively refines a workflow multiple times before arriving at the final workflow. In such an exploratory setting, it is valuable to…
As open source software (OSS) becomes increasingly mature and popular, there are significant challenges with properly accounting for usability concerns for the diverse end users. Participatory design, where multiple stakeholders collaborate…
The size of Transformer models is growing at an unprecedented pace. It has only taken less than one year to reach trillion-level parameters after the release of GPT-3 (175B). Training such models requires both substantial engineering…
In this paper, we present our vision of differentiable ML pipelines called DiffML to automate the construction of ML pipelines in an end-to-end fashion. The idea is that DiffML allows to jointly train not just the ML model itself but also…
In enterprise organizations, data-driven decision making processes include the use of business intelligence dashboards and collaborative deliberation on communication platforms such as Slack. However, apart from those in data analyst roles,…
Progress in many domains increasingly benefits from our ability to view the systems through a computational lens, i.e., using computational abstractions of the domains; and our ability to acquire, share, integrate, and analyze disparate…
This paper introduces a pipeline to parametrically sample and render multi-task vision datasets from comprehensive 3D scans from the real world. Changing the sampling parameters allows one to "steer" the generated datasets to emphasize…
Embeddings are a powerful way to enrich data-driven machine learning models with the world knowledge of large language models (LLMs). Yet, there is limited evidence on how to design effective LLM-based embedding pipelines for tabular…
In the current landscape of foundation model training, there is a significant reliance on public domain data, which is nearing exhaustion according to recent research. To further scale up, it is crucial to incorporate collaboration among…
Machine learning is disruptive. At the same time, machine learning can only succeed by collaboration among many parties in multiple steps naturally as pipelines in an eco-system, such as collecting data for possible machine learning…
Data is the key to success for any Data-Driven Organization, and managing it is considered the most challenging task. Data Architecture (DA) focuses on describing, collecting, storing, processing, and analyzing the data to meet business…
Reservoir simulations help the energy industry make better decisions by predicting how fluids like oil, gas, water, hydrogen, and carbon dioxide will flow underground. To keep these predictions accurate, engineers often need to update…
This paper introduces H-STREAM, a big stream/data processing pipelines evaluation engine that proposes stream processing operators as micro-services to support the analysis and visualisation of Big Data streams stemming from IoT (Internet…
The number of works addressing the role of energy efficiency in the software development has been increasing recently. But, designers and programmers still complain about the lack of tools that help them to make energy-efficiency decisions.…
Emerging collaborative Peer-to-Peer (P2P) systems require discovery and utilization of diverse, multi-attribute, distributed, and dynamic groups of resources to achieve greater tasks beyond conventional file and processor cycle sharing.…
Investment planning requires knowledge of the financial landscape on a large scale, both in terms of geo-spatial and industry sector distribution. There is plenty of data available, but it is scattered across heterogeneous sources…
Transformer models have emerged as potent solutions to a wide array of multidisciplinary challenges. The deployment of Transformer architectures is significantly hindered by their extensive computational and memory requirements,…
The opacity of machine learning data is a significant threat to ethical data work and intelligible systems. Previous research has addressed this issue by proposing standardized checklists to document datasets. This paper expands that field…
Discovering authoritative links between publications and the datasets that they use can be a labor-intensive process. We introduce a natural language processing pipeline that retrieves and reviews publications for informal references to…