[ML-news] [Jobs] Researcher Position at IBM India Research Lab
DINESH GARG
Dear Colleagues,
We are looking to hire a full-time researcher at IBM India Research Lab in the broad areas of Neuro-Symbolic AI and NLP. Please email me your CV if you are interested in exploring the position.
[Background]
Neuro-Symbolic AI & Reasoning group at IBM India Research Lab is focusing on a wide range of problems that require fusion of neuro-symbolic approaches and cut across multiple areas including NLP, Large Language Models, Knowledge Representation, Deep Learning, Logical Reasoning, Neuro-Symbolic Methods, and Optimization. Some sample problems include Complex Question Answering, Text Generation, Trustworthy Text Generation, AI for Code, Code Translation, etc. We are looking to hire a motivated researcher in any of these areas for a full-time role starting immediately.
[Key Duties]
- Participate in discussion and coordinate with a large-sized team spread across global labs of IBM.
- Define problem, design and implement solutions, fast experimentation, demo, and transfer of technology.
- Publish the work in top venues like NeurIPS, ICML, AAAI, IJCAI, ACL, EMNLP, KDD, ICCV, CVPR, etc.
- Participate enthusiastically in patent filing activities.
- Mentor summer interns and other team members.
[Requirements]
[Education]
- Ph.D. degree in computer science, information science, statistics, neuroscience, cognitive science, or equivalent with a specialization in at least one of the following - AI, NLP, Deep Learning, Reasoning, Knowledge Representation, Large Language Models, Machine Translation.
[Professional and Technical Expertise]
- Strong programming skills in Python.
- Experience with Deep Learning framework PyTorch and NLP libraries such as Spacy, NLTK, etc.
- Experience working with Large Language Models.
- Experience with Deep Learning and NLP techniques: Entity Linking, Relation Linking, Semantic Parsing, Attention Models, KG embeddings, Text Generation.
- Experience in the development or management of software resources/tools, Github, Docker, Kubernetes, REST API Service.
[Competences]
- Good communication and presentation skills.
- Strong technical writing skills.
- Ability to work both independently and within a team.
- Used to work under pressure under strict deadlines
Best Regards,
Dinesh Garg
Research Scientist & Manager,
Neuro-Symbolic AI & Reasoning,
+91 880 009 6802 Mobile
garg....@in.ibm.com
IBM Research
DINESH GARG
Call for Papers: 1st ACM SIGKDD Workshop on Content Understanding and Generation for E-commerce
August 2022 Washington DC, USA
Introduction:
Shopping experience on any e-commerce website is largely driven by the content customers interact with. The content is not just limited to product titles or descriptions but has evolved to include brand stories, how-to-use guides, multimedia advertisements, user reviews, unboxing videos, social media feeds, augmented/virtual reality models etc. Since shoppers actively rely on diverse e-commerce content in their shopping journeys, it is critical for the content to be engaging, high quality, relevant and up to date to deliver high user satisfaction and engagement. However, creating good quality content and structuring it bears significant cognitive load on the creative teams. The large volume of diverse content on e-commerce platforms, and the advances in machine learning, pose unique opportunities for gathering insights through content understanding and applying these insights to generate content for better shopper experience. The first workshop on content understanding and generation for the e-commerce industry aims to bring together researchers from industry and academia on questions surrounding e-commerce content understanding and generation.
The workshop will showcase work and have talks about topics relevant to the following themes:
Content understanding for e-commerce content: Semantic understanding and structuring of multi-modal product content such as finding duplicates, discovering and validating product attributes in cold start settings, online learning for rapidly evolving product inventory or generating novel product representations are some active areas of research. Systems that can effectively reason about e-commerce content can influence design of novel search and recommendation algorithms. They can also inform design decisions of advertisers or creative designers, for instance customizing product summary, highlighting videos or descriptions as per shopping trends, seasonality or different occasions to increase product visibility.
Generation and synthesis of e-commerce content: While there has been a great deal of research on generating text, images or video through these have not necessarily been leveraged for content creation in e-commerce where the challenges are unique. Generative models are more challenging to build for e-commerce content, since their outputs not only have to be realistic but also factually correct (e.g. Q&A and product summary generation), visually appealing and adhere to a brand’s design constraints. Quality of the generated content is yet to be solved at scale. The workshop will provide researchers the opportunity to present and discuss work on generative models (text, video, audio, image), quality/trust assessment, addressing bias in generation and other open problems in e-commerce content generation.
The workshop will solicit contributions related to the theme of supporting generation and curation of content for e-commerce which includes (but is not limited to) the following topics.
- Cold start brand/product summary or promotional video generation
- Novel approaches to generate and evaluate product catalogue, review and comparison summaries
- Multimodal and multi-view content modelling aggregating information from multiple product data sources (including text, images, and video) to support product description or advertisement creation
- Multimodal techniques for matching image and text from product feature/advertisement videos and product description
- Multimodal techniques for understanding affective expressions, including non-visual signals such as audio or speech from product videos and advertisements
- Multimodal action/scene recognition in product feature/advertisement videos across product verticals like sports, healthcare, entertainment or lifestyle
- Domain adaptation methods for understanding user submitted product review texts and images
- Domain adaptation methods for recognition of products in non-natural datasets
- Transfer learning applying models trained on prior datasets on a new type of product or category
- Different approaches and metrics to assess the visual appeal of e-commerce content
- Product catalogue moderation and optimization
- Guided generative models for images, audio and videos based on product script and brand story
- Product description or brand store layout/template generation guided by business metrics such as click through rate (CTR) or revenue
- Generation of interpretable and actionable insights from product catalogue and user submitted reviews for content creators
- Few shot learning approaches to bootstrap generation models for new brands, products or demographics
- Evaluation metrics to assess the quality of generated content
Paper submission Guidelines
We solicit two types of submissions – full papers of 6 pages and short papers of at most 2 pages. The submissions must be in PDF format and use two-columns ACM Conference Proceeding template. Template guidelines are at https://www.acm.org/publications/proceedings-template. Submit your paper through the workshop CMT submission site https://cmt3.research.microsoft.com/EcomGen2022.
Important Dates
- Paper submission deadline: May 26th, 2022
- Notification of decision: June 20th, 2022
- Camera-ready due: July 4th, 2022
Workshop Organizers
- Sumit Negi, Amazon Ads
- Rajdeep Banerjee, Amazon Ads
- Manisha Verma, Amazon Ads
- Pooja A, Amazon Ads
- Mithun DasGupta, Microsoft
- Vinay P. Namboodiri, University of Bath
- Dinesh Garg, IBM Research
- Lydia Chilton, Columbia University
-------------------------
Best Regards,
Dinesh Garg
Senior Research Scientist & Manager,
Neuro-Symbolic AI & Reasoning,
garg....@in.ibm.com
IBM Research