曼彻斯特大学（英国）PhD Position in Foundation models for complex scientific systems申请条件要求-申请方
主办方
曼彻斯特大学（英国）
PhD直招介绍
About the Project
AI_CDT_DecisionMaking
Details
Future scientific facilities will produce more data than humans can inspect, requiring AI to identify important data and when expert intervention is needed. This project uses particle detectors to study how AI can make reliable decisions when data are unfamiliar and labelled examples are limited.
LArTPC detectors produce complex 3D particle data and are used in major experiments including MicroBooNE, SBND, ProtoDUNE and DUNE. Their high-dimensional, rare and changing signals provide an ideal testbed for developing foundation models that recognise uncertainty and the limits of their knowledge.
The Manchester group has pretrained a 91-million-parameter foundation model on nearly one million LArTPC events using self-supervised learning. It improves photon-finding efficiency by 16–22 percentage points and strongly distinguishes real data from simulation (AUC 0.999), revealing detector-level differences missed by conventional methods.
The PhD will develop this into a general framework for decision-making with foundation models across changing scientific instruments.
1. Uncertainty under domain shift. The model is pretrained on real MicroBooNE data, but the labels used to adapt it to each task come from simulation, and all data used so far come from wire-readout detectors. DUNE's first far-detector module uses vertical-drift technology with strip readout. The student will develop calibrated per-event uncertainty and out-of-distribution detection using deep ensembles, Laplace approximations and conformal prediction. The backbone has no concept of wires: readout enters through the input embedding, normalisation and augmentation, so a new detector can be addressed by changing an input adapter while retaining the backbone.
2. Decisions with scarce labels. The student will use uncertainty to decide which events are most valuable for experts to label and when an automated selection should defer to a human. This connects active learning to an explicit decision problem: how much expert attention is needed, and where should it be spent?
3. Humans and machines operating a detector. Shift crews must decide whether incoming data are good and whether intervention is needed. The student will use the learned representation for anomaly detection and data-quality monitoring, with explanations an operator can act on, and study how experts use those recommendations. Available SBND and ProtoDUNE datasets will provide future real-data tests, with a route into DUNE through Pandora in collaboration with Leigh Whitehead.
Pretrained checkpoints exist and the domain-shift tools run on them without further training, so the student can begin a first-year uncertainty study alongside the CDT taught programme. The methodological goal is to understand how foundation models can support reliable decisions in scientific systems that evolve away from their training data. The same problem appears across data-intensive science, including astronomy and autonomous laboratories.
Join us and help grow the UK’s national capability in artificial intelligence, part of the UK’s modern industrial strategy. Everyone is welcome if you are passionate about uncertainty quantification, decision making with humans in the loop and decision making for machine learning systems and their applications in Science and Engineering, and looking to develop your research and innovation skills at the doctoral level to further your career in the AI sector.
Equality, diversity and inclusion is fundamental to the success of The University of Manchester, and is at the heart of all of our activities. We know that diversity strengthens our research community, leading to enhanced research creativity, productivity and quality, and societal and economic impact. (Equality, diversity and inclusion | The University of Manchester( https://www.manchester.ac.uk/connect/jobs/equality-diversity-inclusion/ ))
We actively encourage applicants from diverse career paths and backgrounds and from all sections of the community, regardless of age, disability, ethnicity, gender, gender expression, sexual orientation and transgender status.
曼彻斯特大学（英国） PhD Position in Foundation models for complex scientific systems项目有没有奖学金，是不是全奖Phd招生，下面我们一起看一下【大学名称】Phd的奖学金资助情况
项目资助情况
Funding Notes
Fully funded AI UKRI CDT 4 year program; The Project will be based in University of Manchester.
The UKRI AI CDT in Decision Making for Complex Systems is part of the TechFirst( https://www.gov.uk/guidance/techfirst ) offering stipends of £30,000 for students eligible for home fee status. The start date is September 2027.
曼彻斯特大学（英国）Phd申请条件和要求都有哪些？PhD Position in Foundation models for complex scientific systems项目是不是全奖？有没有奖学金？下面我们一起看一下曼彻斯特大学（英国）申请Phd直招需要具备哪些条件和要求，以及托福、雅思语言成绩要到多少才能申请。
申请要求
Desirable Student Background:
A first or upper-second class degree (or equivalent Master's) in computer science, physics, mathematics, data science or a related discipline, with strong programming skills in Python. Experience with machine learning, for example PyTorch, would be valuable. Prior knowledge of particle physics is not required; the scientific context and detector physics can be learned within the project. The project would particularly suit a student interested in foundation models, uncertainty quantification, representation learning or human–AI decision-making who wants to develop those methods on complex real-world data.
报名方式
招生人信息
Dr JE Evans, Prof LW Whitehead
相关资源
扫描二维码
下载申请方APP
下载申请方APP
扫描二维码
下载申请方APP