The Web Conference (WWW) 2027  ·  Full-Day Workshop

Trust Under Budget

Reliable, Efficient, and Safe (Vision) Large Language Models for Deployment

10–14 May 2027  ·  Dublin, Ireland
Co-located with The Web Conference 2027

Workshop Overview

We propose a full-day workshop on deploying trustworthy (vision-)large language models under real-world resource constraints. In production environments, reliability, efficiency, and safety are not independent objectives — they are jointly bounded by explicit token, compute, memory, latency, energy, and monetary budgets. This workshop introduces a “trust under budget” framework organized around three pillars:

  1. Reliable reasoning under budget — maintaining correctness and faithfulness when inference tokens and compute are constrained;
  2. Efficient inference within deployment budgets — satisfying latency, memory, throughput, and cost limits simultaneously;
  3. Safe deployment after budget-driven optimization — preserving alignment and safety behaviors after quantization, distillation, and compression.

By unifying these challenges under a single budget-aware lens, the workshop bridges the AI research and deployment communities, fostering new methods, benchmarks, and evaluation protocols for (v)LLMs that are not only capable but certifiably trustworthy within industrial constraints.

Objectives

Relevance to The Web Conference 2027

The workshop connects directly to search, recommendation, and advertising under latency and cost constraints; Web agents requiring reliable tool use and prompt-injection defenses; and content generation and moderation at platform scale. It brings together The Web Conference's academic and industry communities in information retrieval, machine learning, and scalable systems. Contributions will link model-level methods to service outcomes such as answer quality, agent reliability, moderation robustness, and serving cost, grounding trustworthy AI research in practical Web deployment.

Tentative Invited Speakers

Specific speaker confirmations to be finalized. (Listed in no particular order.)

Lei Ma
The University of Tokyo, Japan & University of Alberta, Canada
Yarin Gal
University of Oxford
Lianhui Qin
UC San Diego
Hongyu Zhang
Chongqing University & University of Newcastle
Bo Li
University of Illinois Urbana–Champaign
Yinghua Yao
A*STAR Centre for Frontier AI Research

Call for Contributions

Large language models increasingly support search, recommendation, content generation, and autonomous Web agents. Deploying these systems requires more than benchmark accuracy: answers must remain faithful, agents must resist manipulation, and safety checks must fit within serving budgets. As adoption grows, token limits, GPU memory, latency, energy, and cost make reliability, efficiency, and safety inseparable deployment concerns.

This workshop studies these interactions through a “trust under budget” framework. Truncated reasoning can reduce correctness, compression can weaken refusal behavior, and runtime guardrails consume latency. These trade-offs motivate joint evaluation under explicit resource constraints. We welcome theoretical methods, benchmarks, and industrial case studies across three themes:

1. Reliable Reasoning Under Budget

Maintain correctness, faithfulness, and stability with limited tokens and compute:

  • Chain-of-thought verification and hallucination detection
  • Stable latent reasoning under constrained decoding
  • Budget-aware process rewards and self-consistency
  • Adaptive depth, early termination, and reasoning compression
  • Reasoning benchmarks under tightening budgets

2. Efficient Inference Within Deployment Budgets

Meet latency, memory, throughput, and cost requirements:

  • Speculative decoding, efficient attention, and KV-cache compression
  • Quantization-aware serving with quality guarantees
  • Adaptive computation and query-dependent early exit
  • Joint measurement of speed, quality, safety, and cost

3. Safe Deployment After Budget-Driven Optimization

Preserve alignment and robustness after acceleration:

  • Safety preservation after quantization, pruning, and distillation
  • Transfer of refusal behavior to compact models
  • Guardrails and prompt-injection defenses within latency budgets
  • Monitoring unsafe behavior in optimized production models
  • Joint evaluation of efficiency gains and safety preservation

Submission Guidelines

We invite regular papers (up to 8 pages) and short / position papers (up to 4 pages), excluding references, formatted in the ACM style. All submissions receive double-blind peer review via OpenReview. All submissions must be original and not under review elsewhere. Accepted papers will receive poster slots; six will also be selected for contributed talks.

Important Dates

Workshop Papers Submission04 January 2027
Workshop Papers Notification01 February 2027
Final Camera-Ready Papers Deadline16 February 2027
Workshop Date10–14 May 2027 (TBD within conference dates)

All deadlines are 11:59 PM Anywhere on Earth (AoE). Dates will be finalized and announced soon.

Paper Awards

To recognize the strongest contributions, the workshop will present a Best Paper Award and a Best Student Paper Award. Awards are selected by the Program Committee based on technical quality, novelty, and relevance to trustworthy, efficient, and safe (v)LLM deployment under real-world resource budgets. Winners will be invited to present their work in a dedicated in-person oral session.

Best Paper Award

Awarded to the submission that makes the most significant and well-executed contribution across the workshop’s three themes.

Best Student Paper Award

Awarded to the strongest submission whose first author is a student at the time of submission.

Winners will be announced at the workshop. In-person presentation is required to receive the award. Recipients will be listed here after the event.

Rising Star Award Announcement

The Trust Under Budget Language Models (TUBLMs) Rising Star Award recognizes early-career researchers whose work advances reliable, efficient, and safe deployment of (vision-)large language models under explicit resource constraints. Awardees will receive a certificate and be invited to deliver an in-person spotlight talk at the workshop.

Scope and Vision

Modern Web applications increasingly rely on large models that must operate within strict token, compute, memory, latency, energy, and cost budgets. This award highlights early-career researchers who make such systems more reliable, efficient, and safe — and who connect principled methods to real deployment outcomes such as answer quality, agent reliability, moderation robustness, and serving cost. We warmly encourage applications from researchers of underrepresented groups in the AI, ML, Web, and systems communities.

Research Areas of Interest

Eligibility

Application Materials

  1. CV, including a list of publications.
  2. Research statement (up to 2 pages, single column, excluding references) describing accomplishments and future directions.
  3. Two recommendation letters, submitted directly by the referees.

Key Dates

Application Materials DueTBD
Recommendation Letters DueTBD
Awards AnnouncedTBD (at the workshop)

All deadlines are 11:59 PM Anywhere on Earth (AoE). The application portal and dates will be announced soon.

Tentative Schedule

The full-day workshop features six invited talks (30 min each, including Q&A), six contributed talks (15 min each), and a poster session — split across a morning and an afternoon session.

TimeProgramme
09:00–09:05Welcome remarks (Ivor W. Tsang)
Morning Session
09:05–09:35Invited Talk 1
09:35–09:50Contributed Talk 1
09:50–10:20Invited Talk 2
10:20–10:35Contributed Talk 2
10:35–11:05Invited Talk 3
11:05–11:20Contributed Talk 3
11:20–12:20Lunch
12:20–13:30Poster Session
Afternoon Session
13:30–14:00Invited Talk 4
14:00–14:15Contributed Talk 4
14:15–14:45Invited Talk 5
14:45–15:00Contributed Talk 5
15:00–15:30Coffee Break
15:30–16:00Invited Talk 6
16:00–16:15Contributed Talk 6
16:15–16:20Closing remarks

Organizing Committee

Xu Chen
Alibaba Group
Lu Wang
Microsoft Research
Shangding Gu
Shanghai Jiao Tong University
Huangjie Zheng
Apple Research
Jiangchao Yao
Shanghai Jiao Tong University
Alois Knoll
Technical University of Munich
Ivor W. Tsang
A*STAR Centre for Frontier AI Research

Contact

For any questions about the workshop, submissions, or awards, please contact the Organizing Committee:
Email: www_tub_committe@163.com