Berkeley Lab’s machine learning and data science experts will join the global research community across three continents for NeurIPS 2026 (the Fortieth Annual Conference on Neural Information Processing Systems):

• Sydney, Australia: December 6–12, 2026
• Atlanta, Georgia: December 9–13, 2026
• Paris, France: December 9–13, 2026

Throughout the week, Berkeley Lab scientists and engineers will share how they are combining machine learning, high-performance computing, and large-scale scientific data to accelerate scientific discovery. Explore our day-by-day schedule below to see the papers, workshops, and tutorials led by Berkeley Lab teams at NeurIPS 2026.

20Events
4Days
Dec 9, 2026 – Dec 12, 2026Program window

Dates and times shown in Event local time.

Events with times to be announced keep their event local date.

Wednesday, December 9

  1. Poster

    A general kernel framework for non-CND distance measures using -dimensional sparse landmark embeddings

    Speaker
    Marcus Noack , Maher B Alghalayini , Mark Risser
    Date
    Wednesday, December 9, 2026
    Time
    10:00 am–1:00 pm GMT-0500
    Location
    ATLANTA, Georgia World Congress Center, Hall C1
    Full description
    Kernel methods, and Gaussian Processes (GPs) in particular, require a conditionally negative-definite (CND) distance measure to guarantee positive semi-definiteness (PSD) of the kernel matrix — a condition that fails for many natural input spaces, including smooth manifolds and spaces of probability distributions. We propose the Sparse Landmark Embedding (SLE) kernel, which eliminates this requirement entirely. Each input is embedded into a sparse feature vector via compactly supported bump functions centered at all training points; applying any standard PSD kernel in this embedding space yields a kernel that is provably PSD for arbitrary distance measures. The compact support automatically controls embedding sparsity, keeping kernel matrices well-conditioned and computationally tractable despite the high ambient dimension. We provide theoretical guarantees on PSD, sparsity, stability, and universal approximation, and demonstrate, using geodesic and Wasserstein distances, that the SLE kernel matches or substantially exceeds domain-specific baselines in both predictive accuracy and uncertainty quantification.
    Speakers
    • Marcus Noack , Maher B Alghalayini , Mark Risser
  2. Poster

    Do Reasoning LLMs Refuse What They Infer in Long Contexts?

    Speaker
    N. Benjamin Erichson
    Date
    Wednesday, December 9, 2026
    Time
    10:00 am–1:00 pm GMT-0500
    Location
    ATLANTA, Georgia World Congress Center, Hall C1
    Full description
    Long-context LLMs can infer objectives that are not stated explicitly. This capability is useful for reasoning over documents, code, retrieved evidence, and tool traces, but it also creates a safety risk: harmful intent can be distributed across a context and become visible only after the model composes the relevant pieces. Existing safety evaluations mostly test explicit harmful requests, and therefore miss this failure mode. We introduce compositional reasoning attacks, a long-context threat model in which harmful requests are decomposed into semantically incomplete fragments and embedded in long contexts. The final query is neutral; the harmful objective emerges only if the model retrieves the fragments, composes them, and infers the implied goal. We instantiate this setting using AdvBench requests, varying the required reasoning from Direct Retrieval to Single-hop Aggregation, Chain Reasoning, and Multi-hop Deductive Reasoning, and evaluate 15 frontier LLMs on contexts up to 64k tokens. Models usually refuse harmful requests when they are directly retrievable. However, refusal rates drop sharply when the same objectives must be reconstructed compositionally, often with larger failures in longer contexts. Benign reconstruction and fragment-position analyses indicate that these failures are not mainly retrieval errors: models often infer the harmful objective and then comply. Increasing inference-time reasoning improves refusal but remains incomplete and costly. Our results reveal a long-context safety gap: current models are better at refusing harmful requests they see than harmful objectives they infer.
    Speakers
    • N. Benjamin Erichson
  3. Poster

    On Neural Scaling Laws for Weather Emulation through Continual Training

    Speaker
    Shashank Subramanian, Arnur Nigmetov, Dmitriy Morozov, Michael Mahoney
    Date
    Wednesday, December 9, 2026
    Time
    10:00 am–1:00 pm GMT-0500
    Location
    ATLANTA, Georgia World Congress Center, Hall C1
    Full description
    Neural scaling laws, which in some domains can predict the performance of large neural networks as a function of model, data, and compute scale, are the cornerstone of building foundation models in Natural Language Processing and Computer Vision. We study neural scaling in Scientific Machine Learning, focusing on models for weather forecasting. To analyze scaling behavior in as simple a setting as possible, we adopt a minimal, scalable, general-purpose Swin Transformer architecture, and we use continual training with constant learning rates and periodic cooldowns as an efficient training strategy. We show that models trained in this minimalist way follow predictable scaling trends and outperform standard cosine learning rate schedules. Cooldown phases can be re-purposed to improve downstream performance, e.g., enabling accurate multi-step rollouts over longer forecast horizons or sharper predictions through spectral loss adjustments. We also systematically explore a wide range of model and dataset sizes under various compute budgets to construct IsoFLOP curves, and we identify compute-optimal training regimes. Extrapolating these trends to larger scales highlights potential performance limits, demonstrating that neural scaling can serve as an important diagnostic for efficient resource allocation. We open-source our code for reproducibility.
    Speakers
    • Shashank Subramanian, Arnur Nigmetov, Dmitriy Morozov, Michael Mahoney
  4. Poster

    Continuity Laws for Sequential Models

    Speaker
    N. Benjamin Erichson
    Date
    Wednesday, December 9, 2026
    Time
    6:00 pm–9:00 pm GMT-0500
    Location
    ATLANTA, Georgia World Congress Center, Hall C1
    Full description
    Inductive biases influence the behavior and performance of sequential models. In this work, we study an underexplored inductive bias in sequential modeling: continuity in time. We ask a simple question: do models motivated by continuous-time formulations, such as state-space models, actually behave continuously in time, and does this translate into better performance on tasks with continuous temporal structure? To answer this, we formalize model continuity as convergence under temporal refinement, where a model is continuous if its predictions approach an underlying continuous trajectory as the temporal discretization is refined. We show that S4 exhibits stable continuous behavior, whereas S6 (the core of Mamba) can be more sensitive to input amplitude and selective dynamics, despite being derived from a continuous dynamical system. To study whether this distinction matters for learning, we also need a corresponding notion of task continuity. We therefore introduce a metric to quantify the continuity of datasets directly from their temporal structure. Across benchmarks, we find a clear empirical alignment between task continuity, model continuity, and model performance. Beyond an inductive bias, continuity also has practical consequences: we show that it enables a simple temporal subsampling strategy that improves both efficiency and performance.
    Speakers
    • N. Benjamin Erichson
  5. Poster

    Free Decompression with Algebraic Spectral Curves

    Speaker
    Michael Mahoney
    Date
    Wednesday, December 9, 2026
    Time
    6:00 pm–9:00 pm GMT-0500
    Location
    ATLANTA, Georgia World Congress Center, Hall 1-4
    Full description
    Tools from random matrix theory have become central to deep learning theory, using spectral information to provide mechanisms for modeling generalization, robustness, scaling, and failure modes. While often capable of modeling empirical behavior, practical computations are limited by matrix size, often imposing a restriction to models that are too small to be realistic. This motivates the inference of properties of larger models from the behavior of smaller ones. Free decompression (FD) is a recently proposed method for extrapolating spectral information across matrix sizes, but its utility is currently limited by strong assumptions that preclude its implementation on more realistic machine learning (ML) models. We use algebraic spectral curve theory to provide a general FD methodology for spectral densities whose Stieltjes transform satisfies an algebraic relation, a modeling assumption that is more likely to hold in practice. This recasts FD as an evolution along spectral curves which can be readily integrated. Our framework enables the expansion of spectral densities that have multiple or multi-modal bulks, that exist at multiple scales, and that contain atoms, all characteristic of real-world data and popular ML models. We demonstrate the efficacy of our framework on models of interest in modern ML, including Hessian and activation matrices associated with neural networks and large-scale diffusion models.
    Speakers
    • Michael Mahoney

Thursday, December 10

  1. Poster

    Fast and Stable Gradient Approximation for Bilinear Forms of Hermitian Matrix Functions

    Speaker
    Navjot Singh, Sherry Li
    Date
    Thursday, December 10, 2026
    Time
    10:00 am–1:00 pm GMT+1100
    Location
    International Convention Centre Sydney (ICC Sydney), Hall 1-4
    Full description
    Objectives involving bilinear forms (u^\top f(A(\theta))v) for Hermitian (A) arise widely in scientific computing and probabilistic machine learning. For large matrices, Lanczos efficiently approximates these quantities, but differentiating them with respect to (\theta) is challenging. Existing approaches either backpropagate through the Lanczos recurrence, requiring reorthogonalization for stability, or apply Arnoldi to an augmented block matrix of twice the original size. Both introduce extra computation and orthogonalization costs that can limit performance on modern hardware. We propose a forward-only gradient approximation that reuses the Lanczos pass and adds very minimal overhead in most cases. We prove that its error is proportional to the Lanczos residual norm, the same quantity controlling the forward approximation. Whereas a traditional adjoint-based calculation would be unstable without reorthogonalization, the new method appears unconditionally stable in our tests. It is also faster than existing state-of-the-art approaches.
    Speakers
    • Navjot Singh, Sherry Li
  2. Poster

    Online Quantile Omniprediction for Proper Losses

    Speaker
    Michael Mahoney
    Date
    Thursday, December 10, 2026
    Time
    10:00 am–1:00 pm GMT+1100
    Location
    International Convention Centre Sydney (ICC Sydney), Hall 1-4
    Full description
    The task of constructing a predictor that is simultaneously accurate under multiple evaluation metrics has been popularized recently under the name omniprediction. Prior work in this area has largely focused on binary prediction tasks, and existing results for multiclass and continuous outcomes typically suffer from slow error rates. In this paper, we consider forecasts expressed as a set of quantiles over a discrete but otherwise arbitrary set of probability levels. We develop sample-efficient algorithms for constructing forecasts that are simultaneously accurate across all proper losses that satisfy mild regularity conditions, where proper losses are those minimized by predicting the true quantiles. Our algorithms are online and can operate in a dynamic environment while simultaneously taking multiple losses into account, and their forecasts offer the potential to provide diverse end-users with rich information for evaluating risk and informing a range of practical decision-making tasks. As an example, we demonstrate how our procedure can be used to ensemble quantile forecasts of COVID-19 hospitalizations made by experts during the pandemic.
    Speakers
    • Michael Mahoney
  3. Poster

    Stable Resolution-Invariant Emulation in Entropically Controlled Kinetic Methods

    Speaker
    Kareem Hegazy, Jose Antonio Lara Benitez, Michael Mahoney
    Date
    Thursday, December 10, 2026
    Time
    10:00 am–1:00 pm GMT+1100
    Location
    International Convention Centre Sydney (ICC Sydney), Hall 1-4
    Full description
    Resolution invariance is a critical capability for scientific machine learning methods (SciML), however, the ability to sample at any resolution does not guarantee stable emulation at all resolutions. This important distinction underpins one of scientific machine learning's greatest potential: the ability to efficiently emulate complicated and computationally expensive systems on inexpensive coarse grids. Stable numerical solvers are computationally expensive as they attempt to resolve small scales that receive and dissipate spectral energy. Learned coarse-grid forecasters remove the dissipative channel, allowing unresolved frequencies to re-emerge as grid-scale noise (aliasing). We explore stable resolution invariant emulation through physically principled control over entropy dynamics to reinforce the dissipative mechanism within the Neural Discrete Equilibrium (NeurDE) method. Entropically-stabilized NeurDE produces stable long-time rollouts for thousands of forecasting steps on large Reynolds number turbulent flows (Re=50,000) with very coarse grids ( ). Entropically-stabilized NeurDE demonstrates extreme stability at coarse resolutions, outperforming the data-generating stabilized numerical method: a much more challenging baseline than SciML surrogates.
    Speakers
    • Kareem Hegazy, Jose Antonio Lara Benitez, Michael Mahoney
  4. Poster

    Spectral Estimation with Deformed Decompression

    Speaker
    Michael Mahoney
    Date
    Thursday, December 10, 2026
    Time
    5:00 pm–8:00 pm GMT+1100
    Location
    International Convention Centre Sydney (ICC Sydney), Hall 1-4
    Full description
    Sample covariance matrices are fundamental objects in machine learning and statistics, with valuable information encoded into their eigenvalues. When treating modern problems, the scale of these matrices can become prohibitive for tractable computation. This creates a critical need for "spectral decompression": a method for inferring the eigenspectrum of a very large-scale model using only information from its smaller realizations. Existing methods fall into two main categories: those relying on spectral inversion; and those that rely on Free Decompression techniques. On one hand, spectral inversion is a fundamentally ill-posed problem that is notoriously unstable, often failing to yield reliable results in finite-size settings. On the other hand, while Free Decompression was recently proposed to address this scaling, its reliance on the Nica-Speicher framework restricts its use to unitarily invariant ensembles, a symmetry condition that subsampled covariance matrices fail to satisfy. In this work, we introduce Deformed Decompression (DD), a novel framework for the stable, forward-extrapolation of spectral densities. By leveraging properties of the companion matrix, DD recasts the spectral decompression problem as a directed evolution of the Stieltjes transform of the companion. This approach bypasses the instabilities of traditional inversion and provides a robust methodology for scaling covariance matrices and even their generalized variants (including Hessian matrices) that lack unitary invariance. We demonstrate the utility of our framework through experiments on a variety of random matrix examples, showing that DD successfully captures the complex spectral signatures of large-scale systems.
    Speakers
    • Michael Mahoney
  5. Poster

    Do Reasoning LLMs Refuse What They Infer in Long Contexts?

    Speaker
    N. Benjamin Erichson
    Date
    Thursday, December 10, 2026
    Time
    10:00 am–1:00 pm GMT-0500
    Location
    ATLANTA, Georgia World Congress Center, Hall 1-4
    Full description
    Long-context LLMs can infer objectives that are not stated explicitly. This capability is useful for reasoning over documents, code, retrieved evidence, and tool traces, but it also creates a safety risk: harmful intent can be distributed across a context and become visible only after the model composes the relevant pieces. Existing safety evaluations mostly test explicit harmful requests, and therefore miss this failure mode. We introduce compositional reasoning attacks, a long-context threat model in which harmful requests are decomposed into semantically incomplete fragments and embedded in long contexts. The final query is neutral; the harmful objective emerges only if the model retrieves the fragments, composes them, and infers the implied goal. We instantiate this setting using AdvBench requests, varying the required reasoning from Direct Retrieval to Single-hop Aggregation, Chain Reasoning, and Multi-hop Deductive Reasoning, and evaluate 15 frontier LLMs on contexts up to 64k tokens. Models usually refuse harmful requests when they are directly retrievable. However, refusal rates drop sharply when the same objectives must be reconstructed compositionally, often with larger failures in longer contexts. Benign reconstruction and fragment-position analyses indicate that these failures are not mainly retrieval errors: models often infer the harmful objective and then comply. Increasing inference-time reasoning improves refusal but remains incomplete and costly. Our results reveal a long-context safety gap: current models are better at refusing harmful requests they see than harmful objectives they infer.
    Speakers
    • N. Benjamin Erichson
  6. Poster

    MT-JailBench: A Modular Benchmark for Understanding Multi-Turn Jailbreak Attacks

    Speaker
    N. Benjamin Erichson
    Date
    Thursday, December 10, 2026
    Time
    1:30 pm–4:30 pm GMT-0500
    Location
    ATLANTA, Georgia World Congress Center, Hall C1
    Full description
    Multi-turn jailbreaks exploit the ability of large language models to accumulate and act on conversational context. Instead of stating a harmful request directly, an attacker can gradually steer the conversation toward an unsafe answer. Recent methods demonstrate this risk, but they are usually evaluated as black-box pipelines with different budgets, judges, retry rules, and strategy generation procedures. As a result, it is often unclear whether reported gains reflect stronger attack mechanisms or different experimental conditions. We introduce MT-JailBench, a modular evaluation framework for benchmarking multi-turn jailbreaks under fixed conditions. MT-JailBench implements each attack as five interacting modules: evaluation function, attack strategy, prompt generation, prompt refinement, and flow control. This design enables fair comparison across attack methods and component-wise analysis of what drives attack success. Using MT-JailBench, we find that resource budgets and evaluation functions are major confounders: controlling turns, retries, interactions, sampled strategies, and judges substantially change the ranking of attacks. At the component level, prompt generation accounts for most performance variation, while refinement and flow control provide moderate gains. We also find that explicit dynamic strategy generation is not always necessary; stochastic sampling from a fixed strategy can rival more elaborate diversification mechanisms. Finally, recomposing the best components yields a strong attack configuration that outperforms its source attacks and generalizes across diverse target LLMs. MT-JailBench therefore provides a modular framework for comparing multi-turn jailbreaks, understanding the impact of components, and guiding stronger red-teaming evaluations.
    Speakers
    • N. Benjamin Erichson

Friday, December 11

  1. Workshop Poster

    Do AI-for-Science Agents Preserve Model Safety? (Workshop: Third Workshop on Agents in the Wild: Safety, Security, and Beyond)

    Speaker
    N. Benjamin Erichson
    Date
    Friday, December 11, 2026
    Time
    8:00 am–5:00 pm GMT+1100
    Location
    International Convention Centre Sydney (ICC Sydney), Hall 5
    Full description
    AI-for-Science agents are increasingly used to support scientific discovery. These systems combine a base large language model (LLM) with a domain-specific agent harness that connects the model to scientific tools and workflows. While the harness is designed to improve scientific capability, it also changes the system in which the model operates, and its safety behavior need not match that of the base LLM. We ask whether model safety is preserved when an LLM is embedded in an AI-for-Science agent harness. We evaluate four agent harnesses on 50 harmful chemistry requests using three base LLMs and three prompting conditions. Our results show that introducing the harness can significantly increase harmful compliance, including for direct harmful requests. Safety measured on a base LLM therefore cannot be assumed to be preserved in the composed agent. Thus, as AI-for-Science agents become more capable and autonomous, the community needs to evaluate and align the composed agent itself using domain-specific safety benchmarks, rather than assuming that safety properties of the base LLM will carry over.
    Speakers
    • N. Benjamin Erichson
  2. Workshop Poster

    The Dose Makes the Poison: Exposure-Aware Evaluation of Generative Drug Design (Workshop: AI for Drug Discovery: Bridging the Translation Gap)

    Speaker
    N. Benjamin Erichson
    Date
    Friday, December 11, 2026
    Time
    8:30 am–5:00 pm GMT+1100
    Location
    International Convention Centre Sydney (ICC Sydney), MR C3.3
    Full description

    NeurIPS 2026 has not supplied a description for this event.

    Speakers
    • N. Benjamin Erichson
  3. Workshop

    Foundation Models for Temporal Systems (FMTS)

    Speaker
    N. Benjamin Erichson
    Date
    Friday, December 11, 2026
    Time
    9:00 am–5:30 pm GMT+1100
    Location
    SYDNEY, TBD
    Full description
    Bringing forecasting, simulation, multimodal temporal data, and reliability into one research agenda for models of evolving real-world systems.
    Speakers
    • N. Benjamin Erichson
  4. Workshop Poster

    Duhamel Neural Operator for Learning Phase-Field Dynamics (Workshop: Representations for the Physical Sciences)

    Speaker
    Yingheng Tang
    Date
    Friday, December 11, 2026
    Time
    8:30 am–6:15 pm GMT+0100
    Location
    PARIS, Palais des Congrès, Room 352A
    Full description
    Phase-field simulators are used to predict how polarization fields evolve, a task that has been challenging due to their high computational cost. Machine learning surrogates have quicker inference times, but tend to struggle in approximating the amplification of incubation noise. We introduce the Duhamel Neural Operator (Duhamel-NO), a neural operator architecture consisting of a Fourier state-space model and a next-step U-Shaped Neural Operator, whose update is built on Duhamel’s principle applied to the Time-Dependent Ginzburg–Landau (TDGL) theory. Duhamel-NO reproduces the reference solver’s (FERROX) trajectories up to three orders of magnitude faster while exhibiting the most physically consistent predictions and significantly lower error than the best of seven tuned baselines, achieving a near-perfect correlation zero-shot generalization on unseen materials, initializations, and free energy landscapes. Additionally, we construct an open benchmark for ferroelectric phase-field dynamics.
    Speakers
    • Yingheng Tang
  5. Poster

    Learning to Discover Iterative Spectral Algorithms

    Speaker
    Michael Mahoney
    Date
    Friday, December 11, 2026
    Time
    2:30 pm–4:30 pm GMT+0100
    Location
    PARIS, Palais des Congrès, Room 352A
    Full description
    We introduce AutoSpec, a neural network framework for discovering iterative spectral algorithms for large-scale numerical linear algebra and numerical optimization. Our self-supervised models adapt to input operators using coarse spectral information (e.g., eigenvalue estimates and residual norms), and predict recurrence coefficients for computing or applying a matrix polynomial tailored to a downstream task. The effectiveness of AutoSpec relies on three ingredients: an architecture whose inference pass implements short, executable numerical linear algebra recurrences; efficient training on small synthetic problems with transfer to large-scale real-world operators; and task-defined objectives that enforce the desired approximation or preconditioning behavior across the range of spectral profiles represented in the training set. We apply AutoSpec to discovering algorithms for representative tasks on spd matrices: accelerating matrix function approximation; accelerating sparse linear solvers; and spectral filtering/preconditioning for eigenvalue computations. On real-world matrices, the learned procedures deliver orders-of-magnitude improvements in accuracy and/or reductions in iteration count, relative to basic baselines. We find clear connections to classical theory: the induced polynomials may exhibit equioscillation behavior characteristic of Chebyshev polynomial approximation.
    Speakers
    • Michael Mahoney
  6. Poster

    Asymmetric Phase Coding Audio Watermarking

    Speaker
    Fengchen Liu, Zhong Wang
    Date
    Friday, December 11, 2026
    Time
    1:30 pm–4:30 pm GMT-0500
    Location
    ATLANTA, Georgia World Congress Center, Hall C1
    Full description
    The proliferation of deepfake audio challenges voice-based authentication systems; passive forensic detectors are sensitive to evolving generative models and to real-world channel distortions. We propose Asymmetric Phase Coding (APC), a training-free cryptographic signing layer for audio, designed as a compact and auditable provenance primitive that can stand alone or be stacked with learned watermarks. APC combines Ed25519 digital signatures (EdDSA, FIPS 186-5; 64-byte signatures) with Reed–Solomon error correction, pseudo-random STFT phase-bin selection, and a redundant quantization-index-modulation (QIM) code on log-magnitude differences of adjacent bin pairs, yielding a compact, non-repudiable, blind-extractable watermark. We evaluate APC on 1,000 LibriSpeech test-clean clips (10 s each, 44.1 kHz) under eight attack configurations – identity, 10% end-cropping, 20% end-cropping, 8 kHz low-pass, 16 kHz round-trip resampling, FLAC re-encoding, MP3 at 128 kbps, and OGG-Vorbis at 128 kbps – and achieve cryptographic verification rates between 97.5% and 98.3% on every condition at mean PESQ=3.02 and tens-of-milliseconds CPU latency. We explicitly compare APC against recent neural baselines (AudioSeal, WavMark, SilentCipher), detail the threat model (forgery resistance vs. erasure), characterize the dataset, define all metrics, quantify an adaptive white-box erasure attack, and release code, keys, and metadata for reproducibility.
    Speakers
    • Fengchen Liu, Zhong Wang
  7. Poster

    Continuity Laws for Sequential Models

    Speaker
    N. Benjamin Erichson
    Date
    Friday, December 11, 2026
    Time
    4:30 pm–7:30 pm GMT-0500
    Location
    ATLANTA, Georgia World Congress Center, Hall C1
    Full description
    Inductive biases influence the behavior and performance of sequential models. In this work, we study an underexplored inductive bias in sequential modeling: continuity in time. We ask a simple question: do models motivated by continuous-time formulations, such as state-space models, actually behave continuously in time, and does this translate into better performance on tasks with continuous temporal structure? To answer this, we formalize model continuity as convergence under temporal refinement, where a model is continuous if its predictions approach an underlying continuous trajectory as the temporal discretization is refined. We show that S4 exhibits stable continuous behavior, whereas S6 (the core of Mamba) can be more sensitive to input amplitude and selective dynamics, despite being derived from a continuous dynamical system. To study whether this distinction matters for learning, we also need a corresponding notion of task continuity. We therefore introduce a metric to quantify the continuity of datasets directly from their temporal structure. Across benchmarks, we find a clear empirical alignment between task continuity, model continuity, and model performance. Beyond an inductive bias, continuity also has practical consequences: we show that it enables a simple temporal subsampling strategy that improves both efficiency and performance.
    Speakers
    • N. Benjamin Erichson

Saturday, December 12

  1. Workshop Poster

    Choosing the Right Representations: Task-Aware Fusion for Multimodal Materials Language Models (Workshop:AI4Mat: AI for Accelerated Materials Design)

    Speaker
    Alex Morehead, N. Benjamin Erichson
    Date
    Saturday, December 12, 2026
    Time
    Time to be announced
    Location
    SYDNEY, TBD
    Full description
    Multimodal language models for materials science rely on effective mechanisms for incorporating three-dimensional structural representations. MatterChat addresses this challenge using a large bridge module between a pretrained structural encoder and the LLM, together with a separate contrastive alignment stage. We instead ask whether this bridge can adaptively select and combine representations from multiple structural encoders according to the target property. We find that supervised and generative material encoders provide complementary information for property prediction. For Zatom-1, this information is not concentrated in the final layer. Instead, the most informative representations occur at different depths depending on the target property. This motivates Task-Conditioned Attention Fusion (TCAF), a lightweight attention-based bridge that uses task-conditioned latent queries to combine CHGNet features with intermediate Zatom-1 representations. The bridge is trained directly on downstream prediction tasks, allowing representation selection and material-language alignment to be learned jointly without separate contrastive pretraining. Across classification and regression tasks, TCAF outperforms MatterChat, with up to 47.8% lower RMSE, while using substantially fewer trainable parameters and halving the overall training time.
    Speakers
    • Alex Morehead, N. Benjamin Erichson
  2. Competition (Satellite)

    FAIR Universe – Weak Lensing ML Uncertainty Challenge (Phase 2)

    Speaker
    Po-Wen Chang, Wahid Bhimji, Paolo Calafiura, Steven Farrell, and Benjamin Nachman
    Date
    Saturday, December 12, 2026
    Time
    3:30 pm–5:45 pm GMT+0100
    Location
    PARIS, Palais des Congrès, Room TBD
    Full description
    This competition is motivated by the need to quantify and compare the information content that different analysis methods—ranging from classical statistics to ML-based models—can extract from weak lensing maps, while also evaluating their robustness to simulation inaccuracies and observational systematic uncertaintes. The outcomes of this competition are expected to guide the development of next-generation weak lensing analysis pipelines, foster cross-disciplinary collaboration between the astrophysics and machine learning communities, and ultimately improve the reliability of cosmological inference from current and upcoming surveys such as LSST, Euclid, and the Roman Space Telescope. By explicitly addressing simulation-model mismatch and the need to quantify systematic uncertainties, this competition emphasizes scientific robustness and interpretability, aligning with the growing emphasis on trustworthy ML in scientific domains.
    Speakers
    • Po-Wen Chang, Wahid Bhimji, Paolo Calafiura, Steven Farrell, and Benjamin Nachman

Source: NeurIPS 2026. Schedule details may change.

Last edited: November 20, 2026