ACM Multimedia-23 Workshop on Generative AI and Multimodal LLM for Finance (GenAI4Finance)

29-31 October, 2023

Location: Ottawa, Canada

About the GenAI4Finance workshop

Trading and investments in financial markets are useful mechanisms for wealth creation and poverty alleviation. Financial forecasting is an essential task that helps investors make sound investment decisions. It can assist users in better predicting price changes in stocks and currencies along with managing risk to reduce chances of adverse losses and bankruptcies. The world of finance is characterized by millions of people trading numerous assets in complex transactions, giving rise to an overwhelming amount of data that can indicate the future direction of stocks, equities, bonds, currencies, crypto coins, and non-fungible tokens (NFTs). Traditional statistical techniques have been extensively studied for stock market trading in developed economies. However, these techniques hold less relevance with rising automation in trading, the interconnected nature of the economies, and the advent of high-risk, high-reward cryptocurrencies and decentralized financial products like NFTs. Today, the world of finance has been brought ever closer to technology as technological innovations become a core part of financial systems. Hence, there is a need to systematically analyze how advances in Generative Artificial Intelligence and Large Language Modeling can help citizen investors, professional investment companies and novice enthusiasts better understand the financial markets.

Relevance of Multimodal LLMs and Generative AI in Finance

The biggest hurdle in processing unstructured data through LLMs like ChatGPT, GPT-3.5, Multimodal GPT-4, PaLM, and others is their opaque nature that they learn from large internet-sized corpora. Lack of understanding of what, when, and how these models can be utilized for financial applications remains an unsolved problem. While these models have shown tremendous success in NLP and document processing tasks, their study on financial knowledge has been very limited. The difficulty in automatically capturing real-time stock market information by these LLMs impedes their successful application for traders and finance practitioners. The learning abilities of LLMs on streaming news sentiment expressed through different multimedia has been barely touched upon by researchers in finance.
Artificial Intelligence based methods can help bridge that gap between the two domains through inter-sectional studies. Even though there have been scattered studies in understanding these challenges, there is a lack of a unified forum to focus on the umbrella theme of generative AI + multimodal language modeling for financial forecasting. We expect this workshop to attract researchers from various perspectives to foster research on the intersection of multimodal AI and financial forecasting. We also plan to organize one shared task to encourage practitioners and researchers to gain a deeper understanding of the problems and come up with novel solutions.

Topics

This workshop will hold a research track and a shared task. The research track aims to explore recent advances and challenges of Multimodal Generative LLMs for finance. Researchers from artificial intelligence, computer vision, speech processing, natural language processing, data mining, statistics, optimization, and other fields are invited to submit papers on recent advances, resources, tools, and challenges on the broad theme of Multimodal AI for finance. The topics of the workshop include but are not limited to the following:

  • Generative AI applications in finance
  • LLM evaluation for finance
  • Fine-tuning LLMs for finance
  • Pre-training strategies for Financial LLMs
  • Retrieval augmentations for interpretable finance
  • Hallucinations in Financial LLMs and generative AI
  • Machine Learning for Time Series Data
  • Audio-Video processing for Financial Unstructured Data
  • Processing text transcript with audio/video for audio-visual-textual alignment, information extraction, salient event detection, entity linking
  • Conversational dialogue modeling using LLMs
  • Natural Language Processing Applications in finance
  • Named-entity recognition, relationship extraction, ontology learning in financial documents
  • Multimodal financial knowledge discovery
  • Data construction, acquisition, augmentation, feature engineering, and analysis for investment and risk management
  • Bias analysis and mitigation in financial LLMs and datasets
  • Interpretability and explainability for financial Generative AI models and Multimodal LLMs
  • Privacy-preserving AI for finance
  • Video understanding for financial multimedia
  • Vision + language and/or other modalities
  • Time series modeling for financial applications
Important Dates
  • June 1, 2023: ACM MM 2023 Workshop Website Open
  • August 10, 2023: Submission Deadline
  • August 15, 2023: Paper notification
  • August 18, 2023: Camera-Ready Deadline

All deadlines are “anywhere on earth” (UTC-12)

Submission

Authors are invited to submit their unpublished work that represents novel research. The papers should be written in English using the ACM Multimedia-23 author kit and follow the ACM Multimedia-23 formatting guidelines. Authors can also submit supplementary materials, including technical appendices, source codes, datasets, and multimedia appendices. All submissions, including the main paper and its supplementary materials, should be fully anonymized. For more information on formatting and anonymity guidelines, please refer to ACM Multimedia-23 call for papers page. Reviewing: At least two reviewers with matching technical expertise will review each paper.
Each paper should be accompanied by one workshop registration.
Multiple submissions of the same paper to more ACM Multimedia-2023 workshops are forbidden.

All papers will be double-blind peer-reviewed. The workshop accepts both long papers and short papers:

  • Short Paper: Up to 4 pages of content, plus unlimited pages for references and appendix. Upon acceptance, the authors are provided with 1 more page to address the reviewer's comments.

  • Long Paper: Up to 8 pages of content, plus unlimited pages for references and appendix. Upon acceptance, the authors are provided with 1 more page to address the reviewer's comments.

Two reviewers with the same technical expertise will review each paper. Authors of the accepted papers will present their work in either the Oral or Poster session. All accepted papers will appear in the workshop proceedings to be published in ACM Multimedia 2023 proceedings. The authors will keep the copyright of their published papers.

Paper must be submitted using EasyChair.

Contact us: puneetm@umd.edu