Numerical simulation of non-linear partial differential equations plays a crucial role in modeling physical science and engineering phenomena, such as weather, climate, and aerodynamics. Recent Machine Learning (ML) models trained on low-resolution spatio-temporal signals have shown new promises in capturing important dynamics in high-resolution signals, under the condition that the models can effectively recover the missing details. However, this study shows that significant information is often lost in the low-resolution down-sampled features. To address such issues, we propose a new approach, namely Temporal Stencil Modeling (TSM), which combines the strengths of advanced time-series sequence modeling (with the HiPPO features) and state-of-the-art neural PDE solvers (with learnable stencil modeling). TSM aims to recover the lost information from the PDE trajectories and can be regarded as a temporal generalization of classic finite volume methods such as WENO. Our experimental results show that TSM achieves the new state-of-the-art simulation accuracy for 2-D incompressible Navier-Stokes turbulent flows: it significantly outperforms the previously reported best results by 19.9% in terms of the highly-correlated duration time and reduces the inference latency into 80%. We also show a strong generalization ability of the proposed method to various out-of-distribution turbulent flow settings. Our code is available at "https://github.com/Edward-Sun/TSM-PDE".
Unpaired image-to-image translation has broad applications in art, design, and scientific simulations. One early breakthrough was CycleGAN that emphasizes one-to-one mappings between two unpaired image domains via generative-adversarial networks (GAN) coupled with the cycle-consistency constraint, while more recent works promote one-to-many mapping to boost diversity of the translated images. Motivated by scientific simulation and one-to-one needs, this work revisits the classic CycleGAN framework and boosts its performance to outperform more contemporary models without relaxing the cycle-consistency constraint. To achieve this, we equip the generator with a Vision Transformer (ViT) and employ necessary training and regularization techniques. Compared to previous best-performing models, our model performs better and retains a strong correlation between the original and translated image. An accompanying ablation study shows that both the gradient penalty and self-supervised pre-training are crucial to the improvement. To promote reproducibility and open science, the source code, hyperparameter configurations, and pre-trained model are available at https://github.com/LS4GAN/uvcgan.
Dmitrii Torbunov, Yi Huang, Huan-Hsin Tseng et al.
An unpaired image-to-image (I2I) translation technique seeks to find a mapping between two domains of data in a fully unsupervised manner. While initial solutions to the I2I problem were provided by generative adversarial neural networks (GANs), diffusion models (DMs) currently hold the state-of-the-art status on the I2I translation benchmarks in terms of Frechet inception distance (FID). Yet, DMs suffer from limitations, such as not using data from the source domain during the training or maintaining consistency of the source and translated images only via simple pixel-wise errors. This work improves a recent UVCGAN model and equips it with modern advancements in model architectures and training procedures. The resulting revised model significantly outperforms other advanced GAN- and DM-based competitors on a variety of benchmarks. In the case of Male-to-Female translation of CelebA, the model achieves more than 40% improvement in FID score compared to the state-of-the-art results. This work also demonstrates the ineffectiveness of the pixel-wise I2I translation faithfulness metrics and suggests their revision. The code and trained models are available at https://github.com/LS4GAN/uvcgan2
Shuaiwen Leon Song, Bonnie Kruft, Minjia Zhang et al. · microsoft-research
In the upcoming decade, deep learning may revolutionize the natural sciences, enhancing our capacity to model and predict natural occurrences. This could herald a new era of scientific exploration, bringing significant advancements across sectors from drug development to renewable energy. To answer this call, we present DeepSpeed4Science initiative (deepspeed4science.ai) which aims to build unique capabilities through AI system technology innovations to help domain experts to unlock today's biggest science mysteries. By leveraging DeepSpeed's current technology pillars (training, inference and compression) as base technology enablers, DeepSpeed4Science will create a new set of AI system technologies tailored for accelerating scientific discoveries by addressing their unique complexity beyond the common technical approaches used for accelerating generic large language models (LLMs). In this paper, we showcase the early progress we made with DeepSpeed4Science in addressing two of the critical system challenges in structural biology research.
Deciding how to optimally deploy sensors in a large, complex, and spatially extended structure is critical to ensure that the surface pressure field is accurately captured for subsequent analysis and design. In some cases, reconstruction of missing data is required in downstream tasks such as the development of digital twins. This paper presents a data-driven sparse sensor selection algorithm, aiming to provide the most information contents for reconstructing aerodynamic characteristics of wind pressures over tall building structures parsimoniously. The algorithm first fits a set of basis functions to the training data, then applies a computationally efficient QR algorithm that ranks existing pressure sensors in order of importance based on the state reconstruction to this tailored basis. The findings of this study show that the proposed algorithm successfully reconstructs the aerodynamic characteristics of tall buildings from sparse measurement locations, generating stable and optimal solutions across a range of conditions. As a result, this study serves as a promising first step toward leveraging the success of data-driven and machine learning algorithms to supplement traditional genetic algorithms currently used in wind engineering.
There are various sources of ionizing radiation exposure, where medical exposure for radiation therapy or diagnosis is the most common human-made source. Understanding how gene expression is modulated after ionizing radiation exposure and investigating the presence of any dose-dependent gene expression patterns have broad implications for health risks from radiotherapy, medical radiation diagnostic procedures, as well as other environmental exposure. In this paper, we perform a comprehensive pathway-based analysis of gene expression profiles in response to low-dose radiation exposure, in order to examine the potential mechanism of gene regulation underlying such responses. To accomplish this goal, we employ a statistical framework to determine whether a specific group of genes belonging to a known pathway display coordinated expression patterns that are modulated in a manner consistent with the radiation level. Findings in our study suggest that there exist complex yet consistent signatures that reflect the molecular response to radiation exposure, which differ between low-dose and high-dose radiation.
High-energy large-scale particle colliders produce data at high speed in the order of 1 terabytes per second in nuclear physics and petabytes per second in high-energy physics. Developing real-time data compression algorithms to reduce such data at high throughput to fit permanent storage has drawn increasing attention. Specifically, at the newly constructed sPHENIX experiment at the Relativistic Heavy Ion Collider (RHIC), a time projection chamber is used as the main tracking detector, which records particle trajectories in a volume of a three-dimensional (3D) cylinder. The resulting data are usually very sparse with occupancy around 10.8%. Such sparsity presents a challenge to conventional learning-free lossy compression algorithms, such as SZ, ZFP, and MGARD. The 3D convolutional neural network (CNN)-based approach, Bicephalous Convolutional Autoencoder (BCAE), outperforms traditional methods both in compression rate and reconstruction accuracy. BCAE can also utilize the computation power of graphical processing units suitable for deployment in a modern heterogeneous high-performance computing environment. This work introduces two BCAE variants: BCAE++ and BCAE-2D. BCAE++ achieves a 15% better compression ratio and a 77% better reconstruction accuracy measured in mean absolute error compared with BCAE. BCAE-2D treats the radial direction as the channel dimension of an image, resulting in a 3x speedup in compression throughput. In addition, we demonstrate an unbalanced autoencoder with a larger decoder can improve reconstruction accuracy without significantly sacrificing throughput. Lastly, we observe both the BCAE++ and BCAE-2D can benefit more from using half-precision mode in throughput (76-79% increase) without loss in reconstruction accuracy. The source code and links to data and pretrained models can be found at https://github.com/BNL-DAQ-LDRD/NeuralCompression_v2.
Peter Yongho Kim, Junbeom Kwon, Sunghwan Joo et al.
Modeling spatiotemporal brain dynamics from high-dimensional data, such as functional Magnetic Resonance Imaging (fMRI), is a formidable task in neuroscience. Existing approaches for fMRI analysis utilize hand-crafted features, but the process of feature extraction risks losing essential information in fMRI scans. To address this challenge, we present SwiFT (Swin 4D fMRI Transformer), a Swin Transformer architecture that can learn brain dynamics directly from fMRI volumes in a memory and computation-efficient manner. SwiFT achieves this by implementing a 4D window multi-head self-attention mechanism and absolute positional embeddings. We evaluate SwiFT using multiple large-scale resting-state fMRI datasets, including the Human Connectome Project (HCP), Adolescent Brain Cognitive Development (ABCD), and UK Biobank (UKB) datasets, to predict sex, age, and cognitive intelligence. Our experimental outcomes reveal that SwiFT consistently outperforms recent state-of-the-art models. Furthermore, by leveraging its end-to-end learning capability, we show that contrastive loss-based self-supervised pre-training of SwiFT can enhance performance on downstream tasks. Additionally, we employ an explainable AI method to identify the brain regions associated with sex classification. To our knowledge, SwiFT is the first Swin Transformer architecture to process dimensional spatiotemporal brain functional data in an end-to-end fashion. Our work holds substantial potential in facilitating scalable learning of functional brain imaging in neuroscience research by reducing the hurdles associated with applying Transformer models to high-dimensional fMRI.
Background: Identification of the interactions and regulatory relations between biomolecules play pivotal roles in understanding complex biological systems and the mechanisms underlying diverse biological functions. However, the collection of such molecular interactions has heavily relied on expert curation in the past, making it labor-intensive and time-consuming. To mitigate these challenges, we propose leveraging the capabilities of large language models (LLMs) to automate genome-scale extraction of this crucial knowledge. Results: In this study, we investigate the efficacy of various LLMs in addressing biological tasks, such as the recognition of protein interactions, identification of genes linked to pathways affected by low-dose radiation, and the delineation of gene regulatory relationships. Overall, the larger models exhibited superior performance, indicating their potential for specific tasks that involve the extraction of complex interactions among genes and proteins. Although these models possessed detailed information for distinct gene and protein groups, they faced challenges in identifying groups with diverse functions and in recognizing highly correlated gene regulatory relationships. Conclusions: By conducting a comprehensive assessment of the state-of-the-art models using well-established molecular interaction and pathway databases, our study reveals that LLMs can identify genes/proteins associated with pathways of interest and predict their interactions to a certain extent. Furthermore, these models can provide important insights, marking a noteworthy stride toward advancing our understanding of biological systems through AI-assisted knowledge discovery.
We present an extensible method for identifying semantic points to reverse engineer (i.e. extract the values of) data charts, particularly those in scientific articles. Our method uses a point proposal network (akin to region proposal networks for object detection) to directly predict the position of points of interest in a chart, and it is readily extensible to multiple chart types and chart elements. We focus on complex bar charts in the scientific literature, on which our model is able to detect salient points with an accuracy of 0.8705 F1 (@1.5-cell max deviation); it achieves 0.9810 F1 on synthetically-generated charts similar to those used in prior works. We also explore training exclusively on synthetic data with novel augmentations, reaching surprisingly competent performance in this way (0.6621 F1) on real charts with widely varying appearance, and we further demonstrate our unchanged method applied directly to synthetic pie charts (0.8343 F1). Datasets, trained models, and evaluation code are available at https://github.com/BNLNLP/PPN_model.
Rod Rofougaran, Shinjae Yoo, Huan-Hsin Tseng et al.
The preservation of privacy is a critical concern in the implementation of artificial intelligence on sensitive training data. There are several techniques to preserve data privacy but quantum computations are inherently more secure due to the no-cloning theorem, resulting in a most desirable computational platform on top of the potential quantum advantages. There have been prior works in protecting data privacy by Quantum Federated Learning (QFL) and Quantum Differential Privacy (QDP) studied independently. However, to the best of our knowledge, no prior work has addressed both QFL and QDP together yet. Here, we propose to combine these privacy-preserving methods and implement them on the quantum platform, so that we can achieve comprehensive protection against data leakage (QFL) and model inversion attacks (QDP). This implementation promises more efficient and secure artificial intelligence. In this paper, we present a successful implementation of these privacy-preservation methods by performing the binary classification of the Cats vs Dogs dataset. Using our quantum-classical machine learning model, we obtained a test accuracy of over 0.98, while maintaining epsilon values less than 1.3. We show that federated differentially private training is a viable privacy preservation method for quantum machine learning on Noisy Intermediate-Scale Quantum (NISQ) devices.
Fast screening of drug molecules based on the ligand binding affinity is an important step in the drug discovery pipeline. Graph neural fingerprint is a promising method for developing molecular docking surrogates with high throughput and great fidelity. In this study, we built a COVID-19 drug docking dataset of about 300,000 drug candidates on 23 coronavirus protein targets. With this dataset, we trained graph neural fingerprint docking models for high-throughput virtual COVID-19 drug screening. The graph neural fingerprint models yield high prediction accuracy on docking scores with the mean squared error lower than $0.21$ kcal/mol for most of the docking targets, showing significant improvement over conventional circular fingerprint methods. To make the neural fingerprints transferable for unknown targets, we also propose a transferable graph neural fingerprint method trained on multiple targets. With comparable accuracy to target-specific graph neural fingerprint models, the transferable model exhibits superb training and data efficiency. We highlight that the impact of this study extends beyond COVID-19 dataset, as our approach for fast virtual ligand screening can be easily adapted and integrated into a general machine learning-accelerated pipeline to battle future bio-threats.
Domain Generalization (DG) aims to learn a generalizable model on the unseen target domain by only training on the multiple observed source domains. Although a variety of DG methods have focused on extracting domain-invariant features, the domain-specific class-relevant features have attracted attention and been argued to benefit generalization to the unseen target domain. To take into account the class-relevant domain-specific information, in this paper we propose an Information theory iNspired diSentanglement and pURification modEl (INSURE) to explicitly disentangle the latent features to obtain sufficient and compact (necessary) class-relevant feature for generalization to the unseen domain. Specifically, we first propose an information theory inspired loss function to ensure the disentangled class-relevant features contain sufficient class label information and the other disentangled auxiliary feature has sufficient domain information. We further propose a paired purification loss function to let the auxiliary feature discard all the class-relevant information and thus the class-relevant feature will contain sufficient and compact (necessary) class-relevant information. Moreover, instead of using multiple encoders, we propose to use a learnable binary mask as our disentangler to make the disentanglement more efficient and make the disentangled features complementary to each other. We conduct extensive experiments on four widely used DG benchmark datasets including PACS, OfficeHome, TerraIncognita, and DomainNet. The proposed INSURE outperforms the state-of-art methods. We also empirically show that domain-specific class-relevant features are beneficial for domain generalization.
A major concern of deep learning models is the large amount of data that is required to build and train them, much of which is reliant on sensitive and personally identifiable information that is vulnerable to access by third parties. Ideas of using the quantum internet to address this issue have been previously proposed, which would enable fast and completely secure online communications. Previous work has yielded a hybrid quantum-classical transfer learning scheme for classical data and communication with a hub-spoke topology. While quantum communication is secure from eavesdrop attacks and no measurements from quantum to classical translation, due to no cloning theorem, hub-spoke topology is not ideal for quantum communication without quantum memory. Here we seek to improve this model by implementing a decentralized ring topology for the federated learning scheme, where each client is given a portion of the entire dataset and only performs training on that set. We also demonstrate the first successful use of quantum weights for quantum federated learning, which allows us to perform our training entirely in quantum.
The rapid advancement of quantum computing (QC) and machine learning (ML) has given rise to the burgeoning field of quantum machine learning (QML), aiming to capitalize on the strengths of quantum computing to propel ML forward. Despite its promise, crafting effective QML models necessitates profound expertise to strike a delicate balance between model intricacy and feasibility on Noisy Intermediate-Scale Quantum (NISQ) devices. While complex models offer robust representation capabilities, their extensive circuit depth may impede seamless execution on extant noisy quantum platforms. In this paper, we address this quandary of QML model design by employing deep reinforcement learning to explore proficient QML model architectures tailored for designated supervised learning tasks. Specifically, our methodology involves training an RL agent to devise policies that facilitate the discovery of QML models without predetermined ansatz. Furthermore, we integrate an adaptive mechanism to dynamically adjust the learning objectives, fostering continuous improvement in the agent's learning process. Through extensive numerical simulations, we illustrate the efficacy of our approach within the realm of classification tasks. Our proposed method successfully identifies VQC architectures capable of achieving high classification accuracy while minimizing gate depth. This pioneering approach not only advances the study of AI-driven quantum circuit design but also holds significant promise for enhancing performance in the NISQ era.
Hsin-Yi Lin, Huan-Hsin Tseng, Samuel Yen-Chi Chen et al.
Quantum machine learning (QML) has recently made significant advancements in various topics. Despite the successes, the safety and interpretability of QML applications have not been thoroughly investigated. This work proposes using Variational Quantum Circuits (VQCs) for activation mapping to enhance model transparency, introducing the Quantum Gradient Class Activation Map (QGrad-CAM). This hybrid quantum-classical computing framework leverages both quantum and classical strengths and gives access to the derivation of an explicit formula of feature map importance. Experimental results demonstrate significant, fine-grained, class-discriminative visual explanations generated across both image and speech datasets.
While deep neural networks (DNNs) have revolutionized many fields, their fragility to carefully designed adversarial attacks impedes the usage of DNNs in safety-critical applications. In this paper, we strive to explore the robust features which are not affected by the adversarial perturbations, i.e., invariant to the clean image and its adversarial examples, to improve the model's adversarial robustness. Specifically, we propose a feature disentanglement model to segregate the robust features from non-robust features and domain specific features. The extensive experiments on four widely used datasets with different attacks demonstrate that robust features obtained from our model improve the model's adversarial robustness compared to the state-of-the-art approaches. Moreover, the trained domain discriminator is able to identify the domain specific features from the clean images and adversarial examples almost perfectly. This enables adversarial example detection without incurring additional computational costs. With that, we can also specify different classifiers for clean images and adversarial examples, thereby avoiding any drop in clean image accuracy.
Because protein-protein interactions (PPIs) are crucial to understand living systems, harvesting these data is essential to probe disease development and discern gene/protein functions and biological processes. Some curated datasets contain PPI data derived from the literature and other sources (e.g., IntAct, BioGrid, DIP, and HPRD). However, they are far from exhaustive, and their maintenance is a labor-intensive process. On the other hand, machine learning methods to automate PPI knowledge extraction from the scientific literature have been limited by a shortage of appropriate annotated data. This work presents a unified, multi-source PPI corpora with vetted interaction definitions augmented by binary interaction type labels and a Transformer-based deep learning method that exploits entities' relational context information for relation representation to improve relation classification performance. The model's performance is evaluated on four widely studied biomedical relation extraction datasets, as well as this work's target PPI datasets, to observe the effectiveness of the representation to relation extraction tasks in various data. Results show the model outperforms prior state-of-the-art models. The code and data are available at: https://github.com/BNLNLP/PPI-Relation-Extraction
In this paper, we introduce RevisitAffectiveMemory, a novel task designed to reconstruct autobiographical memories through audio-visual generation guided by affect extracted from electroencephalogram (EEG) signals. To support this pioneering task, we present the EEG-AffectiveMemory dataset, which encompasses textual descriptions, visuals, music, and EEG recordings collected during memory recall from nine participants. Furthermore, we propose RYM (Revisit Your Memory), a three-stage framework for generating synchronized audio-visual contents while maintaining dynamic personal memory affect trajectories. Experimental results demonstrate our method successfully decodes individual affect dynamics trajectories from neural signals during memory recall (F1=0.9). Also, our approach faithfully reconstructs affect-contextualized audio-visual memory across all subjects, both qualitatively and quantitatively, with participants reporting strong affective concordance between their recalled memories and the generated content. Especially, contents generated from subject-reported affect dynamics showed higher correlation with participants' reported affect dynamics trajectories (r=0.265, p<.05) and received stronger user preference (preference=56%) compared to those generated from randomly reordered affect dynamics. Our approaches advance affect decoding research and its practical applications in personalized media creation via neural-based affect comprehension. Codes and the dataset are available at https://github.com/ioahKwon/Revisiting-Your-Memory.
Shubha R. Kharel, Prashansa Mukim, Piotr Maj et al.
Extreme edge-AI systems, such as those in readout ASICs for radiation detection, must operate under stringent hardware constraints such as micron-level dimensions, sub-milliwatt power, and nanosecond-scale speed while providing clear accuracy advantages over traditional architectures. Finding ideal solutions means identifying optimal AI and ASIC design choices from a design space that has explosively expanded during the merger of these domains, creating non-trivial couplings which together act upon a small set of solutions as constraints tighten. It is impractical, if not impossible, to manually determine ideal choices among possibilities that easily exceed billions even in small-size problems. Existing methods to bridge this gap have leveraged theoretical understanding of hardware to f architecture search. However, the assumptions made in computing such theoretical metrics are too idealized to provide sufficient guidance during the difficult search for a practical implementation. Meanwhile, theoretical estimates for many other crucial metrics (like delay) do not even exist and are similarly variable, dependent on parameters of the process design kit (PDK). To address these challenges, we present a study that employs intelligent search using multi-objective Bayesian optimization, integrating both neural network search and ASIC synthesis in the loop. This approach provides reliable feedback on the collective impact of all cross-domain design choices. We showcase the effectiveness of our approach by finding several Pareto-optimal design choices for effective and efficient neural networks that perform real-time feature extraction from input pulses within the individual pixels of a readout ASIC.
Neural style transfer (NST) has evolved significantly in recent years. Yet, despite its rapid progress and advancement, existing NST methods either struggle to transfer aesthetic information from a style effectively or suffer from high computational costs and inefficiencies in feature disentanglement due to using pre-trained models. This work proposes a lightweight but effective model, AesFA -- Aesthetic Feature-Aware NST. The primary idea is to decompose the image via its frequencies to better disentangle aesthetic styles from the reference image while training the entire model in an end-to-end manner to exclude pre-trained models at inference completely. To improve the network's ability to extract more distinct representations and further enhance the stylization quality, this work introduces a new aesthetic feature: contrastive loss. Extensive experiments and ablations show the approach not only outperforms recent NST methods in terms of stylization quality, but it also achieves faster inference. Codes are available at https://github.com/Sooyyoungg/AesFA.
While deep neural networks have shown impressive performance in many tasks, they are fragile to carefully designed adversarial attacks. We propose a novel adversarial training-based model by Attention Guided Knowledge Distillation and Bi-directional Metric Learning (AGKD-BML). The attention knowledge is obtained from a weight-fixed model trained on a clean dataset, referred to as a teacher model, and transferred to a model that is under training on adversarial examples (AEs), referred to as a student model. In this way, the student model is able to focus on the correct region, as well as correcting the intermediate features corrupted by AEs to eventually improve the model accuracy. Moreover, to efficiently regularize the representation in feature space, we propose a bidirectional metric learning. Specifically, given a clean image, it is first attacked to its most confusing class to get the forward AE. A clean image in the most confusing class is then randomly picked and attacked back to the original class to get the backward AE. A triplet loss is then used to shorten the representation distance between original image and its AE, while enlarge that between the forward and backward AEs. We conduct extensive adversarial robustness experiments on two widely used datasets with different attacks. Our proposed AGKD-BML model consistently outperforms the state-of-the-art approaches. The code of AGKD-BML will be available at: https://github.com/hongw579/AGKD-BML.
Danny Dongyeop Han, Yonghyeon Gwon, Ahhyun Lucy Lee et al.
Electrophysiology signals such as EEG and iEEG are central to neuroscience, brain-computer interfaces, and clinical applications, yet existing foundation models remain limited in scale despite clear evidence that scaling improves performance. We introduce DIVER-1, a family of EEG and iEEG foundation models trained on the largest and most diverse corpus to date-5.3k hours of iEEG and 54k hours of EEG (1.6M channel-hours from over 17.7k subjects)-and scaled up to 1.82B parameters. We present the first systematic scaling law analysis for this domain, showing that they follow data-constrained scaling laws: for a given amount of data and compute, smaller models trained for extended epochs consistently outperform larger models trained briefly. This behavior contrasts with prior electrophysiology foundation models that emphasized model size over training duration. To achieve strong performance, we also design architectural innovations including any-variate attention, sliding temporal conditional positional encoding, and multi-domain reconstruction. DIVER-1 iEEG and EEG models each achieve state-of-the-art performance on their respective benchmarks, establishing a concrete guidelines for efficient scaling and resource allocation in electrophysiology foundation model development.
Mahdi Chehimi, Samuel Yen-Chi Chen, Walid Saad et al.
Quantum federated learning (QFL) can facilitate collaborative learning across multiple clients using quantum machine learning (QML) models, while preserving data privacy. Although recent advances in QFL span different tasks like classification while leveraging several data types, no prior work has focused on developing a QFL framework that utilizes temporal data to approximate functions useful to analyze the performance of distributed quantum sensing networks. In this paper, a novel QFL framework that is the first to integrate quantum long short-term memory (QLSTM) models with temporal data is proposed. The proposed federated QLSTM (FedQLSTM) framework is exploited for performing the task of function approximation. In this regard, three key use cases are presented: Bessel function approximation, sinusoidal delayed quantum feedback control function approximation, and Struve function approximation. Simulation results confirm that, for all considered use cases, the proposed FedQLSTM framework achieves a faster convergence rate under one local training epoch, minimizing the overall computations, and saving 25-33% of the number of communication rounds needed until convergence compared to an FL framework with classical LSTM models.
Samuel Yen-Chi Chen, Huan-Hsin Tseng, Hsin-Yi Lin et al.
The rapid progress in quantum computing (QC) and machine learning (ML) has attracted growing attention, prompting extensive research into quantum machine learning (QML) algorithms to solve diverse and complex problems. Designing high-performance QML models demands expert-level proficiency, which remains a significant obstacle to the broader adoption of QML. A few major hurdles include crafting effective data encoding techniques and parameterized quantum circuits, both of which are crucial to the performance of QML models. Additionally, the measurement phase is frequently overlooked-most current QML models rely on pre-defined measurement protocols that often fail to account for the specific problem being addressed. We introduce a novel approach that makes the observable of the quantum system-specifically, the Hermitian matrix-learnable. Our method features an end-to-end differentiable learning framework, where the parameterized observable is trained alongside the ordinary quantum circuit parameters simultaneously. Using numerical simulations, we show that the proposed method can identify observables for variational quantum circuits that lead to improved outcomes, such as higher classification accuracy, thereby boosting the overall performance of QML models.
Huan-Hsin Tseng, Hsin-Yi Lin, Samuel Yen-Chi Chen et al.
This work analyzes transfer learning of the Variational Quantum Circuit (VQC). Our framework begins with a pretrained VQC configured in one domain and calculates the transition of 1-parameter unitary subgroups required for a new domain. A formalism is established to investigate the adaptability and capability of a VQC under the analysis of loss bounds. Our theory observes knowledge transfer in VQCs and provides a heuristic interpretation for the mechanism. An analytical fine-tuning method is derived to attain the optimal transition for adaptations of similar domains.
Large-scale scientific collaborations like ATLAS, Belle II, CMS, DUNE, and others involve hundreds of research institutes and thousands of researchers spread across the globe. These experiments generate petabytes of data, with volumes soon expected to reach exabytes. Consequently, there is a growing need for computation, including structured data processing from raw data to consumer-ready derived data, extensive Monte Carlo simulation campaigns, and a wide range of end-user analysis. To manage these computational and storage demands, centralized workflow and data management systems are implemented. However, decisions regarding data placement and payload allocation are often made disjointly and via heuristic means. A significant obstacle in adopting more effective heuristic or AI-driven solutions is the absence of a quick and reliable introspective dynamic model to evaluate and refine alternative approaches. In this study, we aim to develop such an interactive system using real-world data. By examining job execution records from the PanDA workflow management system, we have pinpointed key performance indicators such as queuing time, error rate, and the extent of remote data access. The dataset includes five months of activity. Additionally, we are creating a generative AI model to simulate time series of payloads, which incorporate visible features like category, event count, and submitting group, as well as hidden features like the total computational load-derived from existing PanDA records and computing site capabilities. These hidden features, which are not visible to job allocators, whether heuristic or AI-driven, influence factors such as queuing times and data movement.
Uncertainty quantification (UQ) methods play an important role in reducing errors in weather forecasting. Conventional approaches in UQ for weather forecasting rely on generating an ensemble of forecasts from physics-based simulations to estimate the uncertainty. However, it is computationally expensive to generate many forecasts to predict real-time extreme weather events. Evidential Deep Learning (EDL) is an uncertainty-aware deep learning approach designed to provide confidence about its predictions using only one forecast. It treats learning as an evidence acquisition process where more evidence is interpreted as increased predictive confidence. We apply EDL to storm forecasting using real-world weather datasets and compare its performance with traditional methods. Our findings indicate that EDL not only reduces computational overhead but also enhances predictive uncertainty. This method opens up novel opportunities in research areas such as climate risk assessment, where quantifying the uncertainty about future climate is crucial.
William Watkins, Heehwan Wang, Sangyoon Bae et al.
The utility of machine learning has rapidly expanded in the last two decades and presents an ethical challenge. Papernot et. al. developed a technique, known as Private Aggregation of Teacher Ensembles (PATE) to enable federated learning in which multiple teacher models are trained on disjoint datasets. This study is the first to apply PATE to an ensemble of quantum neural networks (QNN) to pave a new way of ensuring privacy in quantum machine learning (QML) models.
Samuel Yen-Chi Chen, Huan-Hsin Tseng, Hsin-Yi Lin et al.
The rapid advancements in quantum computing (QC) and machine learning (ML) have sparked significant interest, driving extensive exploration of quantum machine learning (QML) algorithms to address a wide range of complex challenges. The development of high-performance QML models requires expert-level expertise, presenting a key challenge to the widespread adoption of QML. Critical obstacles include the design of effective data encoding strategies and parameterized quantum circuits, both of which are vital for the performance of QML models. Furthermore, the measurement process is often neglected-most existing QML models employ predefined measurement schemes that may not align with the specific requirements of the targeted problem. We propose an innovative framework that renders the observable of a quantum system-specifically, the Hermitian matrix-trainable. This approach employs an end-to-end differentiable learning framework, enabling simultaneous optimization of the neural network used to program the parameterized observables and the standard quantum circuit parameters. Notably, the quantum observable parameters are dynamically programmed by the neural network, allowing the observables to adapt in real time based on the input data stream. Through numerical simulations, we demonstrate that the proposed method effectively programs observables dynamically within variational quantum circuits, achieving superior results compared to existing approaches. Notably, it delivers enhanced performance metrics, such as higher classification accuracy, thereby significantly improving the overall effectiveness of QML models.
Music style transfer enables personalized music creation by blending the structure of a source with the stylistic attributes of a reference. Existing text-conditioned and diffusion-based approaches show promise but often require paired datasets, extensive training, or detailed annotations. We present Stylus, a training-free framework that repurposes a pre-trained Stable Diffusion model for music style transfer in the mel-spectrogram domain. Stylus manipulates self-attention by injecting style key-value features while preserving source queries to maintain musical structure. To improve fidelity, we introduce a phase-preserving reconstruction strategy that avoids artifacts from Griffin-Lim reconstruction, and we adopt classifier-free-guidance-inspired control for adjustable stylization and multi-style blending. In extensive evaluations, Stylus outperforms state-of-the-art baselines, achieving 34.1% higher content preservation and 25.7% better perceptual quality without any additional training.
Junghoon Justin Park, Jiook Cha, Samuel Yen-Chi Chen et al.
Practical Quantum Machine Learning (QML) is challenged by noise, limited scalability, and poor trainability in Variational Quantum Circuits (VQCs) on current hardware. We propose a multi-chip ensemble VQC framework that systematically overcomes these hurdles. By partitioning high-dimensional computations across ensembles of smaller, independently operating quantum chips and leveraging controlled inter-chip entanglement boundaries, our approach demonstrably mitigates barren plateaus, enhances generalization, and uniquely reduces both quantum error bias and variance simultaneously without additional mitigation overhead. This allows for robust processing of large-scale data, as validated on standard benchmarks (MNIST, FashionMNIST, CIFAR-10) and a real-world PhysioNet EEG dataset, aligning with emerging modular quantum hardware and paving the way for more scalable QML.
Junghoon Justin Park, Huan-Hsin Tseng, Shinjae Yoo et al.
Quantum reinforcement learning (QRL) promises compact function approximators with access to vast Hilbert spaces, but its practical progress is slowed by NISQ-era constraints such as limited qubits and noise accumulation. We introduce a multi-chip ensemble framework using multiple small Quantum Convolutional Neural Networks (QCNNs) to overcome these constraints. Our approach partitions complex, high-dimensional observations from the Super Mario Bros environment across independent quantum circuits, then classically aggregates their outputs within a Double Deep Q-Network (DDQN) framework. This modular architecture enables QRL in complex environments previously inaccessible to quantum agents, achieving superior performance and learning stability compared to classical baselines and single-chip quantum models. The multi-chip ensemble demonstrates enhanced scalability by reducing information loss from dimensionality reduction while remaining implementable on near-term quantum hardware, providing a practical pathway for applying QRL to real-world problems.
Spatiotemporal learning is challenging due to the intricate interplay between spatial and temporal dependencies, the high dimensionality of the data, and scalability constraints. These challenges are further amplified in scientific domains, where data is often irregularly distributed (e.g., missing values from sensor failures) and high-volume (e.g., high-fidelity simulations), posing additional computational and modeling difficulties. In this paper, we present SCENT, a novel framework for scalable and continuity-informed spatiotemporal representation learning. SCENT unifies interpolation, reconstruction, and forecasting within a single architecture. Built on a transformer-based encoder-processor-decoder backbone, SCENT introduces learnable queries to enhance generalization and a query-wise cross-attention mechanism to effectively capture multi-scale dependencies. To ensure scalability in both data size and model complexity, we incorporate a sparse attention mechanism, enabling flexible output representations and efficient evaluation at arbitrary resolutions. We validate SCENT through extensive simulations and real-world experiments, demonstrating state-of-the-art performance across multiple challenging tasks while achieving superior scalability.
Estimating causal effects from spatiotemporal observational data is essential in public health, environmental science, and policy evaluation, where randomized experiments are often infeasible. Existing approaches, however, either rely on strong structural assumptions or fail to handle key challenges such as interference, spatial confounding, temporal carryover, and time-varying confounding -- where covariates are influenced by past treatments and, in turn, affect future ones. We introduce GST-UNet (G-computation Spatio-Temporal UNet), a theoretically grounded neural framework that combines a U-Net-based spatiotemporal encoder with regression-based iterative G-computation to estimate location-specific potential outcomes under complex intervention sequences. GST-UNet explicitly adjusts for time-varying confounders and captures non-linear spatial and temporal dependencies, enabling valid causal inference from a single observed trajectory in data-scarce settings. We validate its effectiveness in synthetic experiments and in a real-world analysis of wildfire smoke exposure and respiratory hospitalizations during the 2018 California Camp Fire. Together, these results position GST-UNet as a principled and ready-to-use framework for spatiotemporal causal inference, advancing reliable estimation in policy-relevant and scientific domains.
Wei Xu, Derek Freeman DeSantis, Xihaier Luo et al.
Learning a continuous and reliable representation of physical fields from sparse sampling is challenging and it affects diverse scientific disciplines. In a recent work, we present a novel model called MMGN (Multiplicative and Modulated Gabor Network) with implicit neural networks. In this work, we design additional studies leveraging explainability methods to complement the previous experiments and further enhance the understanding of latent representations generated by the model. The adopted methods are general enough to be leveraged for any latent space inspection. Preliminary results demonstrate the contextual information incorporated in the latent representations and their impact on the model performance. As a work in progress, we will continue to verify our findings and develop novel explainability approaches.
Vijayalakshmi Saravanan, Perry Siehien, Shinjae Yoo et al.
Detecting abrupt changes in real-time data streams from scientific simulations presents a challenging task, demanding the deployment of accurate and efficient algorithms. Identifying change points in live data stream involves continuous scrutiny of incoming observations for deviations in their statistical characteristics, particularly in high-volume data scenarios. Maintaining a balance between sudden change detection and minimizing false alarms is vital. Many existing algorithms for this purpose rely on known probability distributions, limiting their feasibility. In this study, we introduce the Kernel-based Cumulative Sum (KCUSUM) algorithm, a non-parametric extension of the traditional Cumulative Sum (CUSUM) method, which has gained prominence for its efficacy in online change point detection under less restrictive conditions. KCUSUM splits itself by comparing incoming samples directly with reference samples and computes a statistic grounded in the Maximum Mean Discrepancy (MMD) non-parametric framework. This approach extends KCUSUM's pertinence to scenarios where only reference samples are available, such as atomic trajectories of proteins in vacuum, facilitating the detection of deviations from the reference sample without prior knowledge of the data's underlying distribution. Furthermore, by harnessing MMD's inherent random-walk structure, we can theoretically analyze KCUSUM's performance across various use cases, including metrics like expected delay and mean runtime to false alarms. Finally, we discuss real-world use cases from scientific simulations such as NWChem CODAR and protein folding data, demonstrating KCUSUM's practical effectiveness in online change point detection.
Tasnuva Chowdhury, Tadashi Maeno, Fatih Furkan Akman et al.
The collaborative efforts of large communities in science experiments, often comprising thousands of global members, reflect a monumental commitment to exploration and discovery. Recently, advanced and complex data processing has gained increasing importance in science experiments. Data processing workflows typically consist of multiple intricate steps, and the precise specification of resource requirements is crucial for each step to allocate optimal resources for effective processing. Estimating resource requirements in advance is challenging due to a wide range of analysis scenarios, varying skill levels among community members, and the continuously increasing spectrum of computing options. One practical approach to mitigate these challenges involves initially processing a subset of each step to measure precise resource utilization from actual processing profiles before completing the entire step. While this two-staged approach enables processing on optimal resources for most of the workflow, it has drawbacks such as initial inaccuracies leading to potential failures and suboptimal resource usage, along with overhead from waiting for initial processing completion, which is critical for fast-turnaround analyses. In this context, our study introduces a novel pipeline of machine learning models within a comprehensive workflow management system, the Production and Distributed Analysis (PanDA) system. These models employ advanced machine learning techniques to predict key resource requirements, overcoming challenges posed by limited upfront knowledge of characteristics at each step. Accurate forecasts of resource requirements enable informed and proactive decision-making in workflow management, enhancing the efficiency of handling diverse, complex workflows across heterogeneous resources.
Junghoon Justin Park, Jungwoo Seo, Sangyoon Bae et al.
Resting-state functional magnetic resonance imaging (fMRI) has emerged as a pivotal tool for revealing intrinsic brain network connectivity and identifying neural biomarkers of neuropsychiatric conditions. However, classical self-attention transformer models--despite their formidable representational power--struggle with quadratic complexity, large parameter counts, and substantial data requirements. To address these barriers, we introduce a Quantum Time-series Transformer, a novel quantum-enhanced transformer architecture leveraging Linear Combination of Unitaries and Quantum Singular Value Transformation. Unlike classical transformers, Quantum Time-series Transformer operates with polylogarithmic computational complexity, markedly reducing training overhead and enabling robust performance even with fewer parameters and limited sample sizes. Empirical evaluation on the largest-scale fMRI datasets from the Adolescent Brain Cognitive Development Study and the UK Biobank demonstrates that Quantum Time-series Transformer achieves comparable or superior predictive performance compared to state-of-the-art classical transformer models, with especially pronounced gains in small-sample scenarios. Interpretability analyses using SHapley Additive exPlanations further reveal that Quantum Time-series Transformer reliably identifies clinically meaningful neural biomarkers of attention-deficit/hyperactivity disorder (ADHD). These findings underscore the promise of quantum-enhanced transformers in advancing computational neuroscience by more efficiently modeling complex spatio-temporal dynamics and improving clinical interpretability.
Large language models have revolutionized artificial intelligence by enabling large, generalizable models trained through self-supervision. This paradigm has inspired the development of scientific foundation models (FMs). However, applying this capability to experimental particle physics is challenging due to the sparse, spatially distributed nature of detector data, which differs dramatically from natural language. This work addresses if an FM for particle physics can scale and generalize across diverse tasks. We introduce a new dataset with more than 11 million particle collision events and a suite of downstream tasks and labeled data for evaluation. We propose a novel self-supervised training method for detector data and demonstrate its neural scalability with models that feature up to 188 million parameters. With frozen weights and task-specific adapters, this FM consistently outperforms baseline models across all downstream tasks. The performance also exhibits robust data-efficient adaptation. Further analysis reveals that the representations extracted by the FM are task-agnostic but can be specialized via a single linear mapping for different downstream tasks.
Fourier Neural Operators (FNOs) offer a principled approach for solving complex partial differential equations (PDEs). However, scaling them to handle more complex PDEs requires increasing the number of Fourier modes, which significantly expands the number of model parameters and makes hyperparameter tuning computationally impractical. To address this, we introduce $μ$Transfer-FNO, a zero-shot hyperparameter transfer technique that enables optimal configurations, tuned on smaller FNOs, to be directly applied to billion-parameter FNOs without additional tuning. Building on the Maximal Update Parametrization ($μ$P) framework, we mathematically derive a parametrization scheme that facilitates the transfer of optimal hyperparameters across models with different numbers of Fourier modes in FNOs, which is validated through extensive experiments on various PDEs. Our empirical study shows that Transfer-FNO reduces computational cost for tuning hyperparameters on large FNOs while maintaining or improving accuracy.
Danny Dongyeop Han, Ahhyun Lucy Lee, Taeyang Lee et al.
Electroencephalography (EEG) is a non-invasive technique widely used in brain-computer interfaces and clinical applications, yet existing EEG foundation models face limitations in modeling spatio-temporal brain dynamics and lack channel permutation equivariance, preventing robust generalization across diverse electrode configurations. To address these challenges, we propose DIVER-0, a novel EEG foundation model that demonstrates how full spatio-temporal attention-rather than segregated spatial or temporal processing-achieves superior performance when properly designed with Rotary Position Embedding (RoPE) for temporal relationships and binary attention biases for channel differentiation. We also introduce Sliding Temporal Conditional Positional Encoding (STCPE), which improves upon existing conditional positional encoding approaches by maintaining both temporal translation equivariance and channel permutation equivariance, enabling robust adaptation to arbitrary electrode configurations unseen during pretraining. Experimental results demonstrate that DIVER-0 achieves competitive performance with only 10% of pretraining data while maintaining consistent results across all channel permutation conditions, validating its effectiveness for cross-dataset generalization and establishing key design principles for handling the inherent heterogeneity of neural recording setups.
Brandon R. Feng, David Keetae Park, Xihaier Luo et al.
Fitting Gaussian Processes (GPs) provides interpretable aleatoric uncertainty quantification for estimation of spatio-temporal fields. Spatio-temporal deep learning models, while scalable, typically assume a simplistic independent covariance matrix for the response, failing to capture the underlying correlation structure. However, spatio-temporal GPs suffer from issues of scalability and various forms of approximation bias resulting from restrictive assumptions of the covariance kernel function. We propose STACI, a novel framework consisting of a variational Bayesian neural network approximation of non-stationary spatio-temporal GP along with a novel spatio-temporal conformal inference algorithm. STACI is highly scalable, taking advantage of GPU training capabilities for neural network models, and provides statistically valid prediction intervals for uncertainty quantification. STACI outperforms competing GPs and deep methods in accurately approximating spatio-temporal processes and we show it easily scales to datasets with millions of observations.
Understanding how the brain's complex nonlinear dynamics give rise to cognitive function remains a central challenge in neuroscience. While brain functional dynamics exhibits scale-free and multifractal properties across temporal scales, conventional neuroimaging analytics assume linearity and stationarity, failing to capture frequency-specific neural computations. Here, we introduce Multi-Band Brain Net (MBBN), the first transformer-based framework to explicitly model frequency-specific spatiotemporal brain dynamics from fMRI. MBBN integrates biologically-grounded frequency decomposition with multi-band self-attention mechanisms, enabling discovery of previously undetectable frequency-dependent network interactions. Trained on 49,673 individuals across three large-scale cohorts (UK Biobank, ABCD, ABIDE), MBBN sets a new state-of-the-art in predicting psychiatric and cognitive outcomes (depression, ADHD, ASD), showing particular strength in classification tasks with up to 52.5\% higher AUROC and provides a novel framework for predicting cognitive intelligence scores. Frequency-resolved analyses uncover disorder-specific signatures: in ADHD, high-frequency fronto-sensorimotor connectivity is attenuated and opercular somatosensory nodes emerge as dynamic hubs; in ASD, orbitofrontal-somatosensory circuits show focal high-frequency disruption together with enhanced ultra-low-frequency coupling between the temporo-parietal junction and prefrontal cortex. By integrating scale-aware neural dynamics with deep learning, MBBN delivers more accurate and interpretable biomarkers, opening avenues for precision psychiatry and developmental neuroscience.
Mesoscale ocean dynamics play a critical role in climate systems, governing heat transport, hurricane genesis, and drought patterns. However, simulating these processes at high resolution remains computationally prohibitive due to their nonlinear, multiscale nature and vast spatiotemporal domains. Implicit neural representations (INRs) reduce the computational costs as resolution-independent surrogates but fail in many-query scenarios (inverse modeling) requiring rapid evaluations across diverse parameters. We present PINROD, a novel framework combining dynamics-aware implicit neural representations with parameterized neural ordinary differential equations to address these limitations. By integrating parametric dependencies into latent dynamics, our method efficiently captures nonlinear oceanic behavior across varying boundary conditions and physical parameters. Experiments on ocean mesoscale activity data show superior accuracy over existing baselines and improved computational efficiency compared to standard numerical simulations.
The human brain is a complex system requiring both macroscopic and microscopic components for comprehensive understanding. However, mapping nonlinear relationships between these scales remains challenging due to technical limitations and the high cost of multimodal Magnetic Resonance Imaging (MRI) acquisition. To address this, we introduce Macro2Micro, a deep learning framework that predicts brain microstructure from macrostructure using a Generative Adversarial Network (GAN). Based on the hypothesis that microscale structural information can be inferred from macroscale structures, Macro2Micro explicitly encodes multiscale brain information into distinct processing branches. To enhance artifact elimination and output quality, we propose a simple yet effective auxiliary discriminator and learning objective. Extensive experiments demonstrated that Macro2Micro faithfully translates T1-weighted MRIs into corresponding Fractional Anisotropy (FA) images, achieving a 6.8\% improvement in the Structural Similarity Index Measure (SSIM) compared to previous methods, while retaining the individual biological characteristics of the brain. With an inference time of less than 0.01 seconds per MR modality translation, Macro2Micro introduces the potential for real-time multimodal rendering in medical and research applications. The code will be made available upon acceptance.
In distributed optimization, the communication of model updates can be a performance bottleneck. Consequently, gradient compression has been proposed as a means of increasing optimization throughput. In general, due to information loss, compression introduces a penalty on the number of iterations needed to reach a solution. In this work, we investigate how the iteration penalty depends on the interaction between compression and problem structure, in the context of non-convex stochastic optimization. We focus on linear schemes, where compression and decompression can be modeled as multiplication with a random matrix. We consider several distributions of matrices, among them Haar-distributed orthogonal matrices and matrices with random Gaussian entries. We find that the impact of compression on convergence can be quantified in terms of a smoothness matrix associated with the objective function, using a norm defined by the compression scheme. The analysis reveals that in certain cases, compression performance is related to low-rank structure or other spectral properties of the problem and our bounds predict that the penalty introduced by compression is significantly reduced compared to worst-case bounds that only consider the compression level, ignoring problem data. We verify the theoretical findings experimentally, including fine-tuning an image classification model.
High-energy large-scale particle colliders generate data at extraordinary rates. Developing real-time high-throughput data compression algorithms to reduce data volume and meet the bandwidth requirement for storage has become increasingly critical. Deep learning is a promising technology that can address this challenging topic. At the newly constructed sPHENIX experiment at the Relativistic Heavy Ion Collider, a Time Projection Chamber (TPC) serves as the main tracking detector, which records three-dimensional particle trajectories in a volume of a gas-filled cylinder. In terms of occupancy, the resulting data flow can be very sparse reaching $10^{-3}$ for proton-proton collisions. Such sparsity presents a challenge to conventional learning-free lossy compression algorithms, such as SZ, ZFP, and MGARD. In contrast, emerging deep learning-based models, particularly those utilizing convolutional neural networks for compression, have outperformed these conventional methods in terms of compression ratios and reconstruction accuracy. However, research on the efficacy of these deep learning models in handling sparse datasets, like those produced in particle colliders, remains limited. Furthermore, most deep learning models do not adapt their processing speeds to data sparsity, which affects efficiency. To address this issue, we propose a novel approach for TPC data compression via key-point identification facilitated by sparse convolution. Our proposed algorithm, BCAE-VS, achieves a $75\%$ improvement in reconstruction accuracy with a $10\%$ increase in compression ratio over the previous state-of-the-art model. Additionally, BCAE-VS manages to achieve these results with a model size over two orders of magnitude smaller. Lastly, we have experimentally verified that as sparsity increases, so does the model's throughput.
Reliably reconstructing physical fields from sparse sensor data is a challenge that frequently arises in many scientific domains. In practice, the process generating the data often is not understood to sufficient accuracy. Therefore, there is a growing interest in using the deep neural network route to address the problem. This work presents a novel approach that learns a continuous representation of the physical field using implicit neural representations (INRs). Specifically, after factorizing spatiotemporal variability into spatial and temporal components using the separation of variables technique, the method learns relevant basis functions from sparsely sampled irregular data points to develop a continuous representation of the data. In experimental evaluations, the proposed model outperforms recent INR methods, offering superior reconstruction quality on simulation data from a state-of-the-art climate model and a second dataset that comprises ultra-high resolution satellite-based sea surface temperature fields.
Xihaier Luo, Balasubramanya T. Nadiga, Yihui Ren et al.
Since model bias and associated initialization shock are serious shortcomings that reduce prediction skills in state-of-the-art decadal climate prediction efforts, we pursue a complementary machine-learning-based approach to climate prediction. The example problem setting we consider consists of predicting natural variability of the North Atlantic sea surface temperature on the interannual timescale in the pre-industrial control simulation of the Community Earth System Model (CESM2). While previous works have considered the use of recurrent networks such as convolutional LSTMs and reservoir computing networks in this and other similar problem settings, we currently focus on the use of feedforward convolutional networks. In particular, we find that a feedforward convolutional network with a Densenet architecture is able to outperform a convolutional LSTM in terms of predictive skill. Next, we go on to consider a probabilistic formulation of the same network based on Stein variational gradient descent and find that in addition to providing useful measures of predictive uncertainty, the probabilistic (Bayesian) version improves on its deterministic counterpart in terms of predictive skill. Finally, we characterize the reliability of the ensemble of ML models obtained in the probabilistic setting by using analysis tools developed in the context of ensemble numerical weather prediction.