Feed aggregator

Comparative Connectomics: What We Learn and What We Miss

Understanding the function of a single neuron requires knowledge of its broader network context. Connectomics—the study of comprehensive neural connection maps, often derived from electron microscopy (EM) datasets—provides that wider context, offering a powerful approach to address fundamental questions in circuit neuroscience.

The release of fully annotated connectomes for the female adult brain1,2, and the male adult ventral nerve cord (VNC) 3,4 now enables the mapping of the ~160,000 neurons within the entire Drosophila nervous system. To comprehend the brain’s influence on motor behaviour, we focused on the Descending Neurons (DNs), a relatively small set of cells that act as a vital bottleneck in conveying essential information from the brain to the VNC . We systematically matched DNs to available Light Microscopy (LM) data to connect these two truncated EM datasets, enabling us to study circuits spanning from sensory neurons in the brain to motor output in the VNC5 . We identified broad sets of DNs that control specific subsets of motor neurons and have compared the full set of DNs in EM between the male and a female adult nerve cord dataset5,6.

Using the Drosophila walking circuit5,7 as an example, I will highlight three key aspects of connectomics research: 1. What can we learn by looking at static map of synaptic connections? 2. How can we study stereotypy and dimorphism with only 2 connectome datasets? 3. What do we miss when looking at connectivity maps?

1Dorkenwald et al. Neuronal wiring diagram of an adult brain. Nature 634, 124–138 (2024). 2Schlegel et al. Whole-brain annotation and multi-connectome cell typing of Drosophila. Nature 634, 139–152 (2024). 3Takemura et al. A Connectome of the Male Drosophila Ventral Nerve Cord. eLife 13:RP97769 (2024). 4Marin et al. Systematic annotation of a complete adult male Drosophila nerve cord connectome reveals principles of functional organisation. eLife 13:RP97766 (2024). 5Stürner and Brooks et al. Comparative connectomics of the descending and ascending neurons of the Drosophila nervous system: stereotypy and sexual dimorphism. bioRxiv 2024.06.04.596633 (2024). 6Azevedo et al. Connectomic reconstruction of a female Drosophila ventral nerve cord. Nature 631, 360–368 (2024). 7Cheong, Eichler and Stürner et al. Transforming descending input into behavior: The organization of premotor circuits in the Drosophila Male Adult Nerve Cord connectomeeLife13:RP96084 (2024).

• Speaker: Tomke Stürner - Drosophila Connectomics Group, Department of Zoology and Neurobiology Division, MRC Laboratory of Molecular Biology
• Wednesday 02 April 2025, 16:00-17:30
• Venue: Max Perutz Lecture Theatre. LMB. Cambridge..
• Series: Cambridge Fly Meetings; organiser: Daniel Sobrido-Cameán.

Add to your calendar or Include in your list

From Machine Learning to Machine Reasoning: Deterministic Neural Syllogistic Reasoning (Part 1).

This talk was recorded https://www.youtube.com/watch?v=9hnM9C4xHeM

In my last talk (https://talks.cam.ac.uk/talk/index/228790), I show four methodological limitations that prevent machine learning systems from reaching the rigour of syllogistic reasoning. They cannot achieve the rigour, not because of insufficient amount of training data, instead, to achieve the rigour, they shall not use training data. What kind of neural networks can be? Neural networks use vector embedding, which is a sphere embedding with zero radius. In this talk, I will show the four limitations can be completely avoided by promoting vector embedding into sphere embedding with non-zero radius and the criterion of achieving deterministic neural reasoning, namely, for any satisfiable reasoning, there is a constant number of M that the neural network shall correctly construct a model within M epochs. I will introduce a novel neural network, Sphere Neural Network (SphNN), which explicitly represents geometric objects, here spheres, and introduces the method of syllogistic reasoning by constructing Euler diagrams in the vector space. Instead of using training data, SphNN uses a neighbourhood transition map to transform the current sphere configuration into the target. SphNN is the first neural network that achieves deterministic human-like syllogistic reasoning in one epoch (M=1).

• Speaker: Tiansi Dong
• Monday 17 March 2025, 17:00-17:45
• Venue: Lecture Theatre 2, Computer Laboratory, William Gates Building.
• Series: Foundation AI; organiser: Pietro Lio.

Add to your calendar or Include in your list

Towards Global Maps of Anthropogenic Threats to Biodiversity and Their Contributions to Species Extinctions

Abstract

Species extinctions are primarily driven by loss of habitat, which is relatively easy to monitor by satellite remote sensing; other anthropogenic threats to biodiversity, like hunting, are much more difficult to observe directly. My PhD project draws on local studies which capture the population effect of some anthropogenic threat, scaling these results using machine learning and remote sensing. In this talk, I will discuss my first attempt at this through quantifying species-specific responses to hunting pressure. I find that machine learning methods can offer marked improvements over (linear) statistical models, which are commonly used in ecology, but model validation must be done carefully to properly contextualise predictive performance. I will preview my plans for integrating these hunting pressure models with the LIFE biodiversity metric framework to express pressure in terms of extinction risk. If there is time, I will also discuss future plans for my PhD.

Bio

Emilio is a PhD student in the Department of Zoology at the University of Cambridge in the Conservation Science Group and the Energy and Environment Group. He is supervised by Andrew Balmford, with co-supervision from Anil Madhavapeddy and Tom Swinfield. He is also part of the AI for Environmental Risks Centre for Doctoral Training, a researcher at the Cambridge Centre for Carbon Credits, and a member of Churchill College. His research focuses on the uses of predictive modeling for biodiversity conservation, with an emphasis on quantifying species-specific responses to human disturbance.

• Speaker: Emilio Luz-Ricca, University of Cambridge
• Friday 21 March 2025, 13:00-13:55
• Venue: FW11, William Gates Building. Zoom link: https://cl-cam-ac-uk.zoom.us/j/4361570789?pwd=Nkl2T3ZLaTZwRm05bzRTOUUxY3Q4QT09&from=addon .
• Series: Energy and Environment Group, Department of CST; organiser: lyr24.

Add to your calendar or Include in your list

Energy Environ. Sci., 2025, Advance Article
DOI: 10.1039/D5EE00159E, Review ArticleBoli Nie, Yanming Meng, Simeng Niu, Longjie Gong, Yufeng Chen, Liujun Guo, Xiang Li, Yan-Chao Wu, Hui-Jing Li, Weiwei Zhang
This review commences with a comprehensive examination of the merits possessed by Janus film based ISE. Subsequently, it elaborates in detail from aspects such as structure, manufacturing materials, and practical applications.
To cite this article before page numbers are assigned, use the DOI form of citation above.
The content of this RSS Feed (c) The Royal Society of Chemistry

Energy Environ. Sci., 2025, Accepted Manuscript
DOI: 10.1039/D4EE05109B, PaperDawei Xi, Zheng Yang, Michael S Emanuel, Panlin Zhao, Michael J. Aziz
Carbon dioxide capture and management are critical technologies for achieving carbon neutrality and mitigating the impacts of global warming. One promising approach for decarbonization involves electrochemical generation of concentrated acid...
The content of this RSS Feed (c) The Royal Society of Chemistry

An Introduction to In-Context Learning

In-context learning (ICL) has emerged as a powerful capability of large language models (LLMs), allowing them to adapt to new tasks without explicit parameter updates. This talk begins with an introduction to meta-learning and neural processes, which lay the foundation for ICL . We then move on to transformer based ICL where the model can be trained from scratch or leverage pre-trained LLMs. To attempt to understand why ICL works, we discuss its connections to Bayesian inference, kernel regression, and gradient descent. Finally, we examine potential safety concerns in ICL , highlighting risks and challenges in reliable AI deployment.

Teams link available upon request (it is sent out on our mailing list, eng-mlg-rcc [at] lists.cam.ac.uk). Sign up to our mailing list for easier reminders via lists.cam.ac.uk.

• Speaker: Juyeon Heo, John Bronskill, University of Cambridge
• Wednesday 19 March 2025, 11:00-12:30
• Venue: Cambridge University Engineering Department, CBL Seminar room BE4-38..
• Series: Machine Learning Reading Group @ CUED; organiser: .

Add to your calendar or Include in your list

Machine Learning for Building-Level Heat Risk Mapping

Title

Machine Learning for Building-Level Heat Risk Mapping

Abstract

Climate change is intensifying the frequency and severity of heat waves, increasing risks to public health and energy systems worldwide. However, many existing heat vulnerability assessments focus primarily on outdoor temperatures, overlooking indoor conditions that directly affect occupants. Although building simulations can reveal the types of buildings whose occupants are most at risk, they rarely pinpoint the exact locations of these vulnerable buildings. In this presentation, I will present a data-driven workflow that locates high-risk buildings and discuss the labeling strategies we have explored for classifying real-world structures using satellite imagery.

Bio

Andrea is a first-year PhD student in the Department of Computer Science and Technology at the University of Cambridge. She is supervised by Prof Srinivasan Keshav. Her research bridges machine learning with civil and environmental engineering, focusing particularly on its applications within the built environment.

• Speaker: Andrea Domiter, University of Cambridge
• Thursday 08 May 2025, 13:00-14:00
• Venue: Room FW11 at the David Attenborough Building. Zoom link: https://cl-cam-ac-uk.zoom.us/j/4361570789?pwd=Nkl2T3ZLaTZwRm05bzRTOUUxY3Q4QT09&from=addon .
• Series: Energy and Environment Group, Department of CST; organiser: lyr24.

Add to your calendar or Include in your list

Transient dynamics under structured perturbations: bridging unstructured and structured pseudospectra

As is known, bounds of the resolvent of a matrix in the right complex half-plane yield bounds of solutions of homogeneous and inhomogeneous linear differential equations with this matrix. We ask two basic questions:

- Up to which size of structured perturbations are the resolvent norms of the perturbed matrices within a given bound in the right complex half-plane?

- For a given size of structured perturbations, what is the smallest common bound for the resolvent norms of the perturbed matrices in the right complex half-plane?

This is considered for general linear structures such as complex or real matrices with a given sparsity pattern or with restricted range and corange, or special classes such as Toeplitz or Hankel matrices. Conceptually, we combine unstructured and structured pseudospectra in a joint pseudospectrum, allowing for the use of resolvent bounds as with unstructured pseudospectra and for structured perturbations as with structured pseudospectra. The above questions are addressed by an algorithm which solves eigenvalue optimization problems via suitably discretized rank-1 matrix differential equations. The talk is based on joint work with Nicola Guglielmi.

• Speaker: Christian Lubich
• Thursday 20 March 2025, 15:00-16:00
• Venue: Centre for Mathematical Sciences, MR14.
• Series: Applied and Computational Analysis; organiser: Matthew Colbrook.

Add to your calendar or Include in your list

Comprehensive analysis of the existing and prospective sodium-ion battery (SIB) systems unveils that a wider application of SIBs should primarily focus on i) regulating the composition, morphology, surface chemistry, safety performance, and production feasibility of mainstream cathodes, ii) enhancing the high-voltage stability, film formation ability, and cost effectiveness of electrolytes, and iii) reducing irreversible capacity loss and interphase growth of hard carbon anodes based on well-understood sodium-ion storage mechanisms.

Abstract

Lithium-ion batteries (LIBs) have been widely adopted in the automotive industry, with an annual global production exceeding 1000 GWh. Despite their success, the escalating demand for LIBs has created concerns on supply chain issues related to key elements, such as lithium, cobalt, and nickel. Sodium-ion batteries (SIBs) are emerging as a promising alternative due to the high abundance and low cost of sodium and other raw materials. Nevertheless, the commercialization of SIBs, particularly for grid storage and automotive applications, faces significant hurdles. This perspective article aims to identify the critical challenges in making SIBs viable from both chemical and techno-economic perspectives. First, a brief comparison of the materials chemistry, working mechanisms, and cost between mainstream LIB systems and prospective SIB systems is provided. The intrinsic challenges of SIBs regarding storage stability, capacity utilization, cycle stability, calendar life, and safe operation of cathode, electrolyte, and anode materials are discussed. Furthermore, issues related to the scalability of material production, materials engineering feasibility, and energy-dense electrode design and fabrication are illustrated. Finally, promising pathways are listed and discussed toward achieving high-energy-density, stable, cost-effective SIBs.

Muscles represent one of the toughest anisotropic soft matters. The appealing features originate from their unique aligned fiber-based structures, which form via a combining process involving two major essentials: absorbing nutrients for matrix growth and mechanical training to strengthen the matrix. In this work, a mechanical training-associated growing strategy is proposed to mimic this process for preparing tough anisotropic hydrogels with tunable sizes and enhanced mechanical properties.

Abstract

Muscles are highly anisotropic, force-bearing issues. They form via a process involving nutrient absorption for matrix growth and mechanical training for matrix toughening, in which cyclic disassembly-reconstruction of muscle fibers plays a critical role in generating strong anisotropic structures. Inspired by this process, a mechanical training-associated growing strategy is developed for preparing tough anisotropic hydrogels. Using anisotropic hydrogels made from polyvinyl alcohol (PVA)/tannic acid (TA) as an example, it is demonstrated that the hydrogels can absorb poly(ethylene glycol) diacrylate (PEGDA) via disassembling their aligned nanofibrillar structures. Incorporation of PEGDA within the hydrogels induces PVA to form crystal domains while subsequent mechanical training can restore the aligned fibrillar structures. Such a combining process results in expansion in materials’ size (≈2 times) and significant enhancement in their mechanical properties (Young's modulus: from 2.4 to 2.85 MPa; ultimate tensile strength: from 8.2 to 14.1 MPa; toughness: from 335 to 465 MJ m−3). With a high energy dissipation efficiency (≈0%), potential applications for these tough and adaptable hydrogels are envisioned in impact-protective materials, surgical sutures, etc.

The charge transfer mechanism and hydrogen evolution reaction photocatalytic reaction mechanism are determined by using transient absorption spectrum, electromagnetic simulation, in situ Raman, and density functional theory theoretical calculation.

Abstract

Plasmonic metal–semiconductor nanocomposites are promising candidates for considerably enhancing the solar-to-hydrogen conversion efficiency of semiconductor-based photocatalysts across the entire solar spectrum. However, the underlying enhancement mechanism remains unclear, and the overall efficiency is still low. Herein, a hollow C@MoS2-Au@CdS nanocomposite photocatalyst is developed to achieve improved photocatalytic hydrogen evolution reaction (HER) across a broad spectral range. Transient absorption spectroscopy experiments and electromagnetic field simulations demonstrate that compared to the treated sample, the untreated sample exhibits a high density of sulfur vacancies. Consequently, under near-field enhancement, photogenerated electrons from CdS and hot electrons generated by intra-band or inter-band transitions of Au nanoparticles are efficiently transferred to the CdS surface, thus significantly improving the HER activity of CdS. Additionally, in situ, Raman spectroscopy provided spectral evidence of S─H intermediate species on the CdS surface during the HER process, which is verified through isotope experiments. Density functional theory simulations identify sulfur atoms in CdS as the catalytic active sites for HER. These findings enhance the understanding of charge transfer mechanisms and HER pathways, offering valuable insights for the design of plasmonic photocatalysts with enhanced efficiency.

Unbalanced charge carrier injections and high densities of non-radiative recombination channels are still major obstacles to advancing high-efficiency blue perovskite light-emitting diodes (LEDs). Here, a deep-HOMO level p-type small molecule, (2-(3,6-dibromo-9H-carbazol-9-yl)ethyl)phosphonic acid, constructs a better-balanced carrier injection due to improved hole and retarded electron injection by spontaneously centered at the bottom and top surfaces of perovskites films, along with modulation of all defects in bulk and at surface of doped films due to the formation of covalent bonds. With this approach a series of blue perovskite LEDs with external quantum efficiencies of up to 24.03% (485 nm), 16.61% (476 nm), and 8.55% (467 nm) is designed.

Abstract

Unbalanced charge carrier injections and high densities of non-radiative recombination channels are still major obstacles to advancing high-efficiency blue perovskite light-emitting diodes (LEDs). Here, a deep-HOMO level p-type small molecule, (2-(3,6-dibromo-9H-carbazol-9-yl)ethyl)phosphonic acid, doped in blue perovskites for building a better-balanced injection and controlling over defects is demonstrated. During the perovskite film deposition process, most small molecules are extruded from the precursor solution to the bottom and top surfaces of the perovskite films. This unique distribution of molecules can construct a better-balanced carrier injection due to improved hole and retarded electron injection by its suitable energy-level structure, along with modulation of all defects in bulk and at the surface of doped films due to the formation of covalent bonds by its functional moiety. With this approach, a series of blue perovskite LEDs is designed with external quantum efficiencies (EQEs) of up to 24.03% (at a luminance of 113 cd m−2 and emission peak of 485 nm), 16.61% (at a luminance of 51 cd m−2 and emission peak of 476 nm) and 8.55% (at a luminance of 30 cd m−2 and emission perk of 467 nm), and encouraging operational stability.

Catalysts for Hydrogen Production

Abstract not available

• Speaker: James Fidler, IEEF and Chemistry
• Thursday 20 March 2025, 11:30-12:30
• Venue: Open Plan Area, Institute for Energy and Environmental Flows, Madingley Rise CB3 0EZ.
• Series: Institute for Energy and Environmental Flows (IEEF); organiser: Catherine Pearson.

Add to your calendar or Include in your list

Title to be confirmed

Abstract not available

• Speaker: Margarida Rodrigue, Department of Veterinary Medicine
• Friday 25 April 2025, 08:45-10:00
• Venue: LT2.
• Series: Friday Morning Seminars, Dept of Veterinary Medicine; organiser: Fiona Roby.

Add to your calendar or Include in your list

Investigating the relationship between inflammatory markers in peripheral blood and clinical presentation of intervertebral disc extrusions in canids

Ruweena graduated from Cambridge vet school in June 2024 and has stayed on to continue her research into IVDE with the neurology department, specifically looking at the inflammatory markers IL-6, IL-1β, MMP -9 & extracellular vesicles in the blood plasma of dogs presented to the QVSH for IVDE .

• Speaker: Ruweena Perera, Department of Veterinary Medicine
• Friday 02 May 2025, 08:45-10:00
• Venue: LT2.
• Series: Friday Morning Seminars, Dept of Veterinary Medicine; organiser: Fiona Roby.

Add to your calendar or Include in your list

Fighting cancer and fake news: A battle against misinformation

Cancer-related medical misinformation is a wicked problem, deeply embedded in social, cultural, and technical systems. It represents a deliberate and profit-driven phenomenon, perpetuated by bad actors exploiting online platforms and societal vulnerabilities. Cancer misinformation thrives on information asymmetry, where creators hold an informational advantage over their audience. Bad actors exploit this imbalance by distorting facts and concealing critical context, preying on knowledge gaps and fear and uncertainty following a diagnosis. Drawing from signalling theory, we will explore how misinformation creators mimic trustworthy signals like expertise (e.g., impersonating professionals), consensus (e.g., fake reviews), and familiarity (e.g., mimicking reputable formats), manipulating audiences into accepting their claims as credible. These individuals and organisations manipulate trust, emotions, and gaps in knowledge, fostering harmful behaviours and undermining public health efforts. Social media’s monetisation systems incentivise engagement over accuracy, perpetuating a vicious cycle of distrust in conventional medicine. Cancer misinformation leads to devastating outcomes, including delays in treatment, financial exploitation, and diminished trust in healthcare systems. Understanding medical misinformation tactics and the structural mechanisms enabling misinformation is critical to devising effective interventions that address its root causes. This talk explores the roots, proliferation, and impacts of cancer-related misinformation, focusing on its mimicking of trust signals, dissemination through digital ecosystems, and profound consequences for patients and caregivers.

• Speaker: Alice Hutchings (University of Cambridge)
• Tuesday 18 March 2025, 14:00-15:00
• Venue: Webinar & LT2, Computer Laboratory, William Gates Building..
• Series: Computer Laboratory Security Seminar; organiser: Anna Talas.

Add to your calendar or Include in your list

BSU Seminar: "A Nonparanormal Approach to Marginal Inference"

Treatment effects for a novel therapy are typically measured by comparing marginal outcome distributions across study arms. While proper randomisation in randomised trials allows their estimation from observed outcome distributions, covariate adjustment is recommended to increase precision. However, for noncollapsible measures like odds or hazard ratios in logistic or proportional hazards models, conditioning on covariates changes the effect interpretation, and different covariate sets yield incomparable estimates.

In this talk, I introduce a novel nonparanormal model formulation for adjusted marginal inference allowing the estimation of the joint distribution of outcome and covariates. This approach offers not only the marginal treatment effect of interest for a wide range of outcome types but also an overall coefficient of determination and covariate-specific measures of prognostic strength. I will show how this method enhances precision of the marginal, noncollapsible effect – both theoretically and through empirical results.

This will be a free hybrid seminar. To register to attend virtually, please click on the link: https://cam-ac-uk.zoom.us/webinar/register/WN_xwbSLwLITemQeb8b-jk-Fg

• Speaker: Susanne Dandl, University of Zurich
• Tuesday 25 March 2025, 14:00-15:00
• Venue: Large Seminar Room, East Forvie Building, Forvie Site Robinson Way Cambridge CB2 0SR..
• Series: MRC Biostatistics Unit Seminars; organiser: Alison Quenault.

Add to your calendar or Include in your list

a) Schematic of all-silicon photodetectors with in-plane photon trapping structures (IPTS). b) Comparison of peak specific detectivity in IPTS photodetectors with previously reported very long-wavelength infrared photodetectors.

Abstract

Silicon (Si) photonics has been widely explored for many various applications, including optical communication, optoelectronic computing, spectroscopy, and image sensing. As a key component for optoelectronic signal conversion in these applications, Si-based infrared photodetectors have attracted extensive attention. However, achieving all-Si on-chip photodetection in the very long-wavelength infrared (VLWIR) range remains challenging, with broadband enhancement and improved operating temperature being pressing issues that need to be addressed. An all-Si photodetector design is presented using in-plane photon trapping structures (IPTS) to enhance detection efficiency and improve the operating temperature of the photodetector at the VLWIR range. The photodetector achieves a broadband enhancement of 285–575% (across 12–19 µm) and a 31% reduction in dark current. Additionally, it exhibits an impressive peak specific detectivity of 1.9 × 1010 cm Hz1/2 W−1 at 15 µm, operating at a temperature of 40 K. This study introduces a novel all-Si optoelectronic device architecture that offers a promising solution for improving the operating temperature and sensitivity of broadband VLWIR devices, making the whole system more compact and cost-effective.

An in-liquid superspreading space-confined epitaxy approach is proposed to fabricate Pt (II) complexes crystalline films on superamphiphilic surface. By regulation of the diffusion process, precise control over the nucleation and crystal growth is achieved, resulting in the formation of planar crystalline films. These films exhibit multi-signal sensing ability, making them suitable for reaffirmed sensing detector in complex and unstable conditions.

Abstract

Solution-based method is regarded as a promising approach to fabricate large-area, high-quality crystalline films, owing to its low-cost manufacturing and facile features. However, traditional solution-based methods still suffer from random simultaneous nucleation and uncontrollable crystal growth which result in polycrystalline films and coffee-ring effect. Herein, it is proposed that an in-liquid superspreading space-confined epitaxy approach on a superamphiphilic surface to fabricate crystalline films with controllable initial nucleation and crystal morphology. With delicate control of the liquid environment, concentration, and superspreading space-confined solvent film thickness, planar crystalline films with high crystallinity and smooth morphology are obtained. A controllable dewetting crystallization mechanism is proposed, indicating that the diffusion coefficient, regulated by liquid environment, can control the dewetting process during crystallization. With the balance of solvent diffusion and solute precipitation in crystallization, the ordered in-plane and out-of-plane molecular stacking is achieved. And the as-prepared planar Pt(II) complex crystalline film exhibits multi-signal sensing ability, which can be further used to fabricate the reaffirmed sensing detector for precise gas sensing in complex and unstable conditions. This study demonstrates a facile approach for crystalline film fabrication with controllable nucleation and morphology in a liquid environment, which holds promising applications in the construction of oxygen or water-sensitive organic/inorganic devices.

This study employs multi-objective kernel-based Bayesian optimization to efficiently screen ionic polymer electrolytes (IPEs) for high-voltage, fast-charging lithium metal batteries (LMBs). By examining just 2.8% of the complex chemical space, it achieves multi-objective optimization for ionic conductivity, electrochemical stability, and capacity in IPEs. These findings highlight the pivotal role of machine learning in expediting material discovery for advanced batteries.

Abstract

Designing ionic polymer electrolytes (IPEs) for high-voltage and fast-charging lithium batteries involves searching in a highly complex and discrete chemical space. Traditional material discovery processes struggle with this complexity due to high costs and long evaluation time. A kernel-based Bayesian optimization is described to complete the multi-objective optimization by considering ionic conductivity, electrochemical stability, and discharge capacity simultaneously. According to a recommender based on a union set of acquisition functions, promising IPEs through three iterations with only 2.8% of the chemical space is targeted. The achieved lithium metal batteries exhibit promising performance with ultrahigh cutoff voltage with NCM811 (LiNi0.8Co0.1Mn0.1O2, 4.8 V) and LNMO (LiNi0.5Mn1.5O4, 4.92 V). To further extend the versatility of IPEs and diminish the high cost associated with the glove-box environment, an aqueous and high-voltage lithium-ion battery is developed by introducing water molecules in IPEs coupled with Li4Ti5O12||LiMn2O4, a strong hydrogen bonding network formed between the rigid-rod polyelectrolyte and the embedded water molecules, which effectively suppresses the water reactivity, meanwhile boosting the ionic conductivity. This work reveals an innovative multi-objective optimization that effectively handles multi-targets and discontinuous parameter space, offering critical insights to address complex challenges in material discovery and property optimization for advanced and versatile lithium batteries.