The deployment of automated pavement defect detection is often hindered by poor cross-domain generalization. Supervised detectors achieve strong in-domain accuracy but require costly re-annotation for new environments, while standard self-supervised methods capture generic features and remain vulnerable to domain shift. We propose PROBE, a self-supervised framework that visually probes target domains without labels. PROBE introduces a Self-supervised Prompt Enhancement Module (SPEM), which derives defect-aware prompts from unlabeled target data to guide a frozen ViT backbone, and a Domain-Aware Prompt Alignment (DAPA) objective, which aligns prompt-conditioned source and target representations. Experiments on four challenging benchmarks show that PROBE consistently outperforms strong supervised, self-supervised, and adaptation baselines, achieving robust zero-shot transfer, improved resilience to domain variations, and high data efficiency in few-shot adaptation. These results highlight self-supervised prompting as a practical direction for building scalable and adaptive visual inspection systems. Source code is publicly available: https://github.com/xixiaouab/PROBE/tree/main

Rapid growth of high-dimensional datasets in fields such as single-cell RNA sequencing and spatial genomics has led to unprecedented opportunities for scientific discovery, but it also presents unique computational and statistical challenges. Traditional methods struggle with geometry-aware data generation, interpolation along meaningful trajectories, and transporting populations via feasible paths. To address these issues, we introduce Geometry-Aware Generative Autoencoder (GAGA), a novel framework that combines extensible manifold learning with generative modeling. GAGA constructs a neural network embedding space that respects the intrinsic geometries discovered by manifold learning and learns a novel warped Riemannian metric on the data space. This warped metric is derived from both the points on the data manifold and negative samples off the manifold, allowing it to characterize a meaningful geometry across the entire latent space. Using this metric, GAGA can uniformly sample points on the manifold, generate points along geodesics, and interpolate between populations across the learned manifold. GAGA shows competitive performance in simulated and real-world datasets, including a 30% improvement over SOTA in single-cell population-level trajectory inference.

In many data-driven applications, higher-order relationships among multiple objects are essential in capturing complex interactions. Hypergraphs, which generalize graphs by allowing edges to connect any number of nodes, provide a flexible and powerful framework for modeling such higher-order relationships. In this work, we introduce hypergraph diffusion wavelets and describe their favorable spectral and spatial properties. We demonstrate their utility for biomedical discovery in spatially resolved transcriptomics by applying the method to represent disease-relevant cellular niches for Alzheimer's disease.

Neuroscience employs diverse neuroimaging techniques, each offering distinct insights into brain activity, from electrophysiological recordings such as EEG, which have high temporal resolution, to hemodynamic modalities such as fMRI, which have increased spatial precision. However, integrating these heterogeneous data sources remains a challenge, which limits a comprehensive understanding of brain function. We present the Spatiotemporal Alignment of Multimodal Brain Activity (SAMBA) framework, which bridges the spatial and temporal resolution gaps across modalities by learning a unified latent space free of modality-specific biases. SAMBA introduces a novel attention-based wavelet decomposition for spectral filtering of electrophysiological recordings, graph attention networks to model functional connectivity between functional brain units, and recurrent layers to capture temporal autocorrelations in brain signal. We show that the training of SAMBA, aside from achieving translation, also learns a rich representation of brain information processing. We showcase this classify external stimuli driving brain activity from the representation learned in hidden layers of SAMBA, paving the way for broad downstream applications in neuroscience research and clinical contexts.

The proliferation of digital microscopy images, driven by advances in automated whole slide scanning, presents significant opportunities for biomedical research and clinical diagnostics. However, accurately annotating densely packed information in these images remains a major challenge. To address this, we introduce DiffKillR, a novel framework that reframes cell annotation as the combination of archetype matching and image registration tasks. DiffKillR employs two complementary neural networks: one that learns a diffeomorphism-invariant feature space for robust cell matching and another that computes the precise warping field between cells for annotation mapping. Using a small set of annotated archetypes, DiffKillR efficiently propagates annotations across large microscopy images, reducing the need for extensive manual labeling. More importantly, it is suitable for any type of pixel-level annotation. We will discuss the theoretical properties of DiffKillR and validate it on three microscopy tasks, demonstrating its advantages over existing supervised, semi-supervised, and unsupervised methods.

Advances in medical imaging technologies have enabled the collection of longitudinal images, which involve repeated scanning of the same patients over time, to monitor disease progression. However, predictive modeling of such data remains challenging due to high dimensionality, irregular sampling, and data sparsity. To address these issues, we propose ImageFlowNet, a novel model designed to forecast disease trajectories from initial images while preserving spatial details. ImageFlowNet first learns multiscale joint representation spaces across patients and time points, then optimizes deterministic or stochastic flow fields within these spaces using a position-parameterized neural ODE/SDE framework. The model leverages a UNet architecture to create robust multiscale representations and mitigates data scarcity by combining knowledge from all patients. We provide theoretical insights that support our formulation of ODEs, and motivate our regularizations involving high-level visual features, latent space organization, and trajectory smoothness. We validate ImageFlowNet on three longitudinal medical image datasets depicting progression in geographic atrophy, multiple sclerosis, and glioblastoma, demonstrating its ability to effectively forecast disease progression and outperform existing methods. Our contributions include the development of ImageFlowNet, its theoretical underpinnings, and empirical validation on real-world datasets.

The rapidly growing field of single-cell transcriptomic sequencing (scRNAseq) presents challenges for data analysis due to its massive datasets. A common method in manifold learning consists in hypothesizing that datasets lie on a lower dimensional manifold. This allows to study the geometry of point clouds by extracting meaningful descriptors like curvature. In this work, we will present Adaptive Local PCA (AdaL-PCA), a data-driven method for accurately estimating various notions of intrinsic curvature on data manifolds, in particular principal curvatures for surfaces. The model relies on local PCA to estimate the tangent spaces. The evaluation of AdaL-PCA on sampled surfaces shows state-of-the-art results. Combined with a PHATE embedding, the model applied to single-cell RNA sequencing data allows us to identify key variations in the cellular differentiation.

Frontier large language models (LLMs) have shown great success in text modeling and generation tasks across domains. However, natural language exhibits inherent semantic hierarchies and nuanced geometric structure, which current LLMs do not capture completely owing to their reliance on Euclidean operations such as dot-products and norms. Furthermore, recent studies have shown that not respecting the underlying geometry of token embeddings leads to training instabilities and degradation of generative capabilities. These findings suggest that shifting to non-Euclidean geometries can better align language models with the underlying geometry of text. We thus propose to operate fully in $\textit{Hyperbolic space}$, known for its expansive, scale-free, and low-distortion properties. To this end, we introduce $\textbf{HELM}$, a family of $\textbf{H}$yp$\textbf{E}$rbolic Large $\textbf{L}$anguage $\textbf{M}$odels, offering a geometric rethinking of the Transformer-based LLM that addresses the representational inflexibility, missing set of necessary operations, and poor scalability of existing hyperbolic LMs. We additionally introduce a $\textbf{Mi}$xture-of-$\textbf{C}$urvature $\textbf{E}$xperts model, $\textbf{HELM-MiCE}$, where each expert operates in a distinct curvature space to encode more fine-grained geometric structure from text, as well as a dense model, $\textbf{HELM-D}$. For $\textbf{HELM-MiCE}$, we further develop hyperbolic Multi-Head Latent Attention ($\textbf{HMLA}$) for efficient, reduced-KV-cache training and inference. For both models, we further develop essential hyperbolic equivalents of rotary positional encodings and root mean square normalization. We are the first to train fully hyperbolic LLMs at billion-parameter scale, and evaluate them on well-known benchmarks such as MMLU and ARC, spanning STEM problem-solving, general knowledge, and commonsense reasoning. Our results show consistent gains from our $\textbf{HELM}$ architectures – up to 4\% – over popular Euclidean architectures used in LLaMA and DeepSeek with superior semantic hierarchy modeling capabilities, highlighting the efficacy and enhanced reasoning afforded by hyperbolic geometry in large-scale language model pretraining.

In this paper, we propose HiPoNet, an end-to-end differentiable neural network for regression, classification, and representation learning on high-dimensional point clouds. Our work is motivated by single-cell data which can have very high-dimensionality --exceeding the capabilities of existing methods for point clouds which are mostly tailored for 3D data. Moreover, modern single-cell and spatial experiments now yield entire cohorts of datasets (i.e., one data set for every patient), necessitating models that can process large, high-dimensional point-clouds at scale. Most current approaches build a single nearest-neighbor graph, discarding important geometric and topological information. In contrast, HiPoNet models the point-cloud as a set of higher-order simplicial complexes, with each particular complex being created using a reweighting of features. This method thus generates multiple constructs corresponding to different views of high-dimensional data, which in biology offers the possibility of disentangling distinct cellular processes. It then employs simplicial wavelet transforms to extract multiscale features, capturing both local and global topology from each view. We show that geometric and topological information is preserved in this framework both theoretically and empirically. We showcase the utility of HiPoNet on point-cloud level tasks, involving classification and regression of entire point-clouds in data cohorts. Experimentally, we find that HiPoNet outperforms other point-cloud and graph-based models on single-cell data. We also apply HiPoNet to spatial transcriptomics datasets using spatial coordinates as one of the views. Overall, HiPoNet offers a robust and scalable solution for high-dimensional data analysis.

Graph neural networks (GNNs) have emerged as a powerful tool for tasks such as node classification and graph classification. However, much less work has been done on signal classification, where the data consists of many functions (referred to as signals) defined on the vertices of a single graph. These tasks require networks designed differently from those designed for traditional GNN tasks. Indeed, traditional GNNs rely on localized low-pass filters, and signals of interest may have intricate multi-frequency behavior and exhibit long range interactions. This motivates us to introduce the BLIS-Net (Bi-Lipschitz Scattering Net), a novel GNN that builds on the previously introduced geometric scattering transform. Our network is able to capture both local and global signal structure and is able to capture both low-frequency and high-frequency information. We make several crucial changes to the original geometric scattering architecture which we prove increase the ability of our network to capture information about the input signal and show that BLIS-Net achieves superior performance on both synthetic and real-world data sets based on traffic flow and fMRI data.

Directed graphs are a natural model for many phenomena, in particular scientific knowledge graphs such as molecular interaction or chemical reaction networks that define cellular signaling relationships. In these situations, source nodes typically have distinct biophysical properties from sinks. Due to their ordered and unidirectional relationships, many such networks also have hierarchical and multiscale structure. However, the majority of methods performing node- and edge-level tasks in machine learning do not take these properties into account, and thus have not been leveraged effectively for scientific tasks such as cellular signaling network inference. We propose a new framework called Directed Scattering Autoencoder (DSAE) which uses a directed version of a geometric scattering transform, combined with the non-linear dimensionality reduction properties of an autoencoder and the geometric properties of the hyperbolic space to learn latent hierarchies. We show this method outperforms numerous others on tasks such as embedding directed graphs and learning cellular signaling networks.

Although data diffusion embeddings are ubiquitous in unsupervised learning and have proven to be a viable technique for uncovering the underlying intrinsic geometry of data, diffusion embeddings are inherently limited due to their discrete nature. To this end, we propose neural FIM, a method for computing the Fisher information metric (FIM) from point cloud data - allowing for a continuous manifold model for the data. Neural FIM creates an extensible metric space from discrete point cloud data such that information from the metric can inform us of manifold characteristics such as volume and geodesics. We demonstrate Neural FIM’s utility in selecting parameters for the PHATE visualization method as well as its ability to obtain information pertaining to local volume illuminating branching points and cluster centers embeddings of a toy dataset and two single-cell datasets of IPSC reprogramming and PBMCs (immune cells).

Diffusion-based manifold learning methods have proven useful in representation learning and dimensionality reduction of modern high dimensional, high throughput, noisy datasets. Such datasets are especially present in fields like biology and physics. While it is thought that these methods preserve underlying manifold structure of data by learning a proxy for geodesic distances, no specific theoretical links have been established. Here, we establish such a link via results in Riemannian geometry explicitly connecting heat diffusion to manifold distances. In this process, we also formulate a more general heat kernel based manifold embedding method that we call heat geodesic embeddings. This novel perspective makes clearer the choices available in manifold learning and denoising. Results show that our method outperforms existing state of the art in preserving ground truth manifold distances, and preserving cluster structure in toy datasets. We also showcase our method on single cell RNA-sequencing datasets with both continuum and cluster structure, where our method enables interpolation of withheld timepoints of data. Finally, we show that parameters of our more general method can be configured to give results similar to PHATE (a state-of-the-art diffusion based manifold learning method) as well as SNE (an attraction/repulsion neighborhood based method that forms the basis of t-SNE).

In modern relational machine learning it is common to encounter large graphs that arise via interactions or similarities between observations in many domains. Further, in many cases the target entities for analysis are actually signals on such graphs. We propose to compare and organize such datasets of graph signals by using an earth mover’s distance (EMD) with a geodesic cost over the underlying graph. Typically, EMD is computed by optimizing over the cost of transporting one probability distribution to another over an underlying metric space. However, this is inefficient when computing the EMD between many signals. Here, we propose an unbalanced graph EMD that efficiently embeds the unbalanced EMD on an underlying graph into an L1space, whose metric we call unbalanced diffusion earth mover’s distance (UDEMD). Next, we show how this gives distances between graph signals that are robust to noise. Finally, we apply this to organizing patients based on clinical notes, embedding cells modeled as signals on a gene graph, and organizing genes modeled as signals over a large cell graph. In each case, we show that UDEMD-based embeddings find accurate distances that are highly efficient compared to other methods.

We introduce a new intrinsic measure of local curvature on point-cloud data called diffusion curvature. Our measure uses the framework of diffusion maps, including the data diffusion operator, to structure point cloud data and define local curvature based on the laziness of a random walk starting at a point or region of the data. We show that this laziness directly relates to volume comparison results from Riemannian geometry. We then extend this scalar curvature notion to an entire quadratic form using neural network estimations based on the diffusion map of point-cloud data. We show applications of both estimations on toy data, single-cell data, and on estimating local Hessian matrices of neural network loss landscapes.

We present a method called Manifold Interpolating Optimal-Transport Flow (MIOFlow) that learns stochastic, continuous population dynamics from static snapshot samples taken at sporadic timepoints. MIOFlow combines dynamic models, manifold learning, and optimal transport by training neural ordinary differential equations (Neural ODE) to interpolate between static population snapshots as penalized by optimal transport with manifold ground distance. Further, we ensure that the flow follows the geometry by operating in the latent space of an autoencoder that we call a geodesic autoencoder (GAE). In GAE the latent space distance between points is regularized to match a novel multiscale geodesic distance on the data manifold that we define. We show that this method is superior to normalizing flows, Schr\"odinger bridges and other generative models that are designed to flow from noise to data in terms of interpolating between populations. Theoretically, we link these trajectories with dynamic optimal transport. We evaluate our method on simulated data with bifurcations and merges, as well as scRNA-seq data from embryoid body differentiation, and acute myeloid leukemia treatment.

A major challenge in embedding or visualizing clinical patient data is the heterogeneity of variable types including continuous lab values, categorical diagnostic codes, as well as missing or incomplete data. In particular, in EHR data, some variables are missing not at random (MNAR) but deliberately not collected and thus are a source of information. For example, lab tests may be deemed necessary for some patients on the basis of suspected diagnosis, but not for others. Here we present the MURAL forest – an unsupervised random forest for representing data with disparate variable types (e.g., categorical, continuous, MNAR). MURAL forests consist of a set of decision trees where node-splitting variables are chosen at random, such that the marginal entropy of all other variables is minimized by the split. This allows us to also split on MNAR variables and discrete variables in a way that is consistent with the continuous variables. The end goal is to learn the MURAL embedding of patients using average tree distances between those patients. These distances can be fed to nonlinear dimensionality reduction method like PHATE to derive visualizable embeddings. While such methods are ubiquitous in continuous-valued datasets (like single cell RNA-sequencing) they have not been used extensively in mixed variable data. We showcase the use of our method on one artificial and two clinical datasets. We show that using our approach, we can visualize and classify data more accurately than competing approaches. Finally, we show that MURAL can also be used to compare cohorts of patients via the recently proposed tree-sliced Wasserstein distances.

We propose a new fast method of measuring distances between large numbers of related high dimensional datasets called the Diffusion Earth Mover’s Distance (EMD). We model the datasets as distributions supported on common data graph that is derived from the affinity matrix computed on the combined data. In such cases where the graph is a discretization of an underlying Riemannian closed manifold, we prove that Diffusion EMD is topologically equivalent to the standard EMD with a geodesic ground distance. Diffusion EMD can be computed in {Õ}(n) time and is more accurate than similarly fast algorithms such as tree-based EMDs. We also show Diffusion EMD is fully differentiable, making it amenable to future uses in gradient-descent frameworks such as deep neural networks. Finally, we demonstrate an application of Diffusion EMD to single cell data collected from 210 COVID-19 patient samples at Yale New Haven Hospital. Here, Diffusion EMD can derive distances between patients on the manifold of cells at least two orders of magnitude faster than equally accurate methods. This distance matrix between patients can be embedded into a higher level patient manifold which uncovers structure and heterogeneity in patients. More generally, Diffusion EMD is applicable to all datasets that are massively collected in parallel in many medical and biological systems.

Traditional Generative Adversarial Networks (GANs) sample from a continuous stochastic noise distribution, and then treating that sample as a constant, optimize the parameters of a generator network on the average loss over the sample. In other words, the generator must take samples from a continuous input distribution, and on average, match a non-continuous target distribution (the training data). The result is that, as the generator transitions from one realistic image to another, some regions of the noise space correspond to relatively high quality images and other regions correspond to relatively low quality images. To avoid these relatively low quality areas, and allow the generator to optimize the shape of its noise distribution in a data-aware way, we present a novel sampling procedure: alternating (a) the traditional approach of optimizing the generator while holding the noise space sample constant, but then also (b) moving points in the noise space to improve the loss, using the direction of the gradient, while holding the generator parameters constant. We demonstrate that this procedure, which we call noise space optimization, improves the overall quality of samples obtained from an identical model without it on a wide array of canonical datasets and training paradigms.

Generative Adversarial Networks (GANs) learn generating functions that map a random noise distribution$Z$to a target data distribution. Usually, little attention is paid to the form of$Z$, resulting in nearly all models assuming$Z$follows a convenient parametric form (almost always$Z\ \sim \ Normal (0,1)$or$Z\ \sim\ Uniform(-1,1))$. However, we observe that image-conditional generators, such as those used in the CycleGAN, produce better quality generation than comparable single domain GANs can currently achieve. This is true even when the images being conditioned upon are substantially different from the target images. We hypothesize these models benefit from input which already has the general structure of images, even if their semantic content is different. As a result, we propose the Canvas GAN: using just a small handful of real images (“canvases”), we create random input for an image-conditional generator by randomly cropping, flipping along either or both axis, coloring, and resizing the canvas. These diverse samples allow the generator to edit an input that already exhibits natural image structure, as opposed to having to generate it from scratch from independent white noise.

Biomolecular graph analysis has recently gained much attention in the emerging field of geometric deep learning. Here we focus on organizing biomolecular graphs in ways that expose meaningful relations and variations between them. We propose a geometric scattering autoencoder (GSAE) network for learning such graph embeddings. Our embedding network first extracts rich graph features using the recently proposed geometric scattering transform. Then, it leverages a semi-supervised variational autoencoder to extract a low-dimensional embedding that retains the information in these features that enable prediction of molecular properties as well as characterize graphs. We show that GSAE organizes RNA graphs both by structure and energy, accurately reflecting b istable R NA s tructures. A lso, the model is generative and can sample new folding trajectories.

It is increasingly common to encounter data in the form of cross-sectional population measurements over time, particularly in biomedical settings. Recent attempts to model individual trajectories from this data use optimal transport to create pairwise matchings between time points. However, these methods cannot model non-linear paths common in many underlying dynamic systems. We establish a link between continuous normalizing flows and dynamic optimal transport to model the expected paths of points over time. Continuous normalizing flows are generally under constrained, as they are allowed to take an arbitrary path from the source to the target distribution. We present \emph{TrajectoryNet}, which controls the continuous paths taken between distributions. We show how this is particularly applicable for studying cellular dynamics in data from single-cell RNA sequencing (scRNA-seq) technologies, and that TrajectoryNet improves upon recently proposed static optimal transport-based models that can be used for interpolating cellular distributions.

While neural networks are powerful approximators used to classify or embed data into lower dimensional spaces, they are often regarded as black boxes with uninterpretable features. Here we propose Graph Spectral Regularization for making hidden layers more interpretable without significantly impacting performance on the primary task. Taking inspiration from spatial organization and localization of neuron activations in biological networks, we use a graph Laplacian penalty to structure the activations within a layer. This penalty encourages activations to be smooth either on a predetermined graph or on a feature-space graph learned from the data via co-activations of a hidden layer of the neural network. We show numerous uses for this additional structure including cluster indication and visualization in biological and image data sets.

Functional magnetic resonance imaging (fMRI) is a crucial technology for gaining insights into cognitive processes in humans. Data amassed from fMRI measurements result in volumetric data sets that vary over time. However, analysing such data presents a challenge due to the large degree of noise and person-to-person variation in how information is represented in the brain. To address this challenge, we present a novel topological approach that encodes each time point in an fMRI data set as a persistence diagram of topological features, i.e. high-dimensional voids present in the data. This representation naturally does not rely on voxel-by-voxel correspondence and is robust towards noise. We show that these time-varying persistence diagrams can be clustered to find meaningful groupings between participants, and that they are also useful in studying within-subject brain state trajectories of subjects performing a particular task. Here, we apply both clustering and trajectory analysis techniques to a group of participants watching the movie 'Partly Cloudy'. We observe significant differences in both brain state trajectories and overall topological activity between adults and children watching the same movie.

We propose a novel framework for combining datasets via alignment of their intrinsic geometry. This alignment can be used to fuse data originating from disparate modalities, or to correct batch effects while preserving intrinsic data structure. Importantly, we do not assume any pointwise correspondence between datasets, but instead rely on correspondence between a (possibly unknown) subset of data features. We leverage this assumption to construct an isometric alignment between the data. This alignment is obtained by relating the expansion of data features in harmonics derived from diffusion operators defined over each dataset. These expansions encode each feature as a function of the data geometry. We use this to relate the diffusion coordinates of each dataset through our assumption of partial feature correspondence. Then, a unified diffusion geometry is constructed over the aligned data, which can also be used to correct the original data measurements. We demonstrate our method on several datasets, showing in particular its effectiveness in biological applications including fusion of single-cell RNA sequencing (scRNA-seq) and single-cell ATAC sequencing (scATAC-seq) data measured on the same population of cells, and removal of batch effect between biological samples.

Big data often has emergent structure that exists at multiple levels of abstraction, which are useful for characterizing complex interactions and dynamics of the observations. Here, we consider multiple levels of abstraction via a multiresolution geometry of data points at different granularities. To construct this geometry we define a time-inhomogemeous diffusion process that effectively condenses data points together to uncover nested groupings at larger and larger granularities. This inhomogeneous process creates a deep cascade of intrinsic low pass filters on the data affinity graph that are applied in sequence to gradually eliminate local variability while adjusting the learned data geometry to increasingly coarser resolutions. We provide visualizations to exhibit our method as a "continuously-hierarchical" clustering with directions of eliminated variation highlighted at each step. The utility of our algorithm is demonstrated via neuronal data condensation, where the constructed multiresolution data geometry uncovers the organization, grouping, and connectivity between neurons.

Archetypal analysis is a data decomposition method that describes each observation in a dataset as a convex combination of ”pure types” or archetypes. These archetypes represent extrema of a data space in which there is a trade-off between features, such as in biology where different combinations of traits provide optimal fitness for different environments. Existing methods for archetypal analysis work well when a linear relationship exists between the feature space and the archetypal space. However, such methods are not applicable to systems where the feature space is generated non-linearly from the combination of archetypes, such as in biological systems or image transformations. Here, we propose a reformulation of the problem such that the goal is to learn a non-linear transformation of the data into a latent archetypal space. To solve this problem, we introduce Archetypal Analysis network (AAnet), which is a deep neural network framework for learning and generating from a latent archetypal representation of data. We demonstrate state-of-the-art recovery of ground-truth archetypes in non-linear data domains, show AAnet can generate from data geometry rather than from data density, and use AAnet to identify biologically meaningful archetypes in single-cell gene expression data.

Interest in image-to-image translation has grown substantially in recent years with the success of unsupervised models based on the cycle-consistency assumption. The achievements of these models have been limited to a particular subset of domains where this assumption yields good results, namely homogeneous domains that are characterized by style or texture differences. We tackle the challenging problem of image-to-image translation where the domains are defined by high-level shapes and contexts, as well as including significant clutter and heterogeneity. For this purpose, we introduce a novel GAN based on preserving intra-domain vector transformations in a latent space learned by a siamese network. The traditional GAN system introduced a discriminator network to guide the generator into generating images in the target domain. To this two-network system we add a third: a siamese network that guides the generator so that each original image shares semantics with its generated version. With this new three-network system, we no longer need to constrain the generators with the ubiquitous cycle-consistency restraint or any other autoencoding regularization. As a result, the generators can learn mappings between more complex domains that differ from each other by more than just style or texture. We demonstrate our model by mapping between high-resolution, arbitrarily chosen classes from the Imagenet dataset completely without pre-processing such as cropping, centering, or filtering unrepresentative images.

Understanding why and how certain neural networks outperform others is key to guiding future development of network architectures and optimization methods. To this end, we introduce a novel visualization algorithm that reveals the internal geometry of such networks: Multislice PHATE (M-PHATE), the first method designed explicitly to visualize how a neural network's hidden representations of data evolve throughout the course of training. We demonstrate that our visualization provides intuitive, detailed summaries of the learning dynamics beyond simple global measures (i.e., validation loss and accuracy), without the need to access validation data. Furthermore, M-PHATE better captures both the dynamics and community structure of the hidden units as compared to visualization based on standard dimensionality reduction methods (e.g., ISOMAP, t-SNE). We demonstrate M-PHATE with two vignettes: continual learning and generalization. In the former, the M-PHATE visualizations display the mechanism of "catastrophic forgetting" which is a major challenge for learning in task-switching contexts. In the latter, our visualizations reveal how increased heterogeneity among hidden units correlates with improved generalization performance. An implementation of M-PHATE, along with scripts to reproduce the figures in this paper, is available at https://github.com/scottgigante/M-PHATE.

It is increasingly common in many types of natural and physical systems (especially biological systems) to have different types of measurements performed on the same underlying system. In such settings, it is important to align the manifolds arising from each measurement in order to integrate such data and gain an improved picture of the system; we tackle this problem using generative adversarial networks (GANs). Recent attempts to use GANs to find correspondences between sets of samples do not explicitly perform proper alignment of manifolds. We present the new Manifold Aligning GAN (MAGAN) that aligns two manifolds such that related points in each measurement space are aligned. We demonstrate applications of MAGAN in single-cell biology in integrating two different measurement types together: cells from the same tissue are measured with both genomic (single-cell RNA-sequencing) and proteomic (mass cytometry) technologies. We show that MAGAN successfully aligns manifolds such that known correlations between measured markers are improved compared to other recently proposed models.

We propose a new type of generative model for high-dimensional data that learns a manifold geometry of the data, rather than density, and can generate points evenly along this manifold. This is in contrast to existing generative models that represent data density, and are strongly affected by noise and other artifacts of data collection. We demonstrate how this approach corrects sampling biases and artifacts, thus improves several downstream data analysis tasks, such as clustering and classification. Finally, we demonstrate that this approach is especially useful in biology where, despite the advent of single-cell technologies, rare subpopulations and gene-interaction relationships are affected by biased sampling. We show that SUGAR can generate hypothetical populations, and it is able to reveal intrinsic patterns and mutual-information relationships between genes on a single-cell RNA sequencing dataset of hematopoiesis.

Eukaryotic cells have complex regulatory systems that sense adversity (e.g. DNA damage, heat shock, external death-induction signals) and respond by invoking programmed cell-death or apoptosis. Cancer cells have evolved the ability to thwart such sensory information and associated regulation. As biologists begin to understand cells in circuit-like terms, we can also begin to derive and simulate druggable targets of cellular networks that cause a cancerous cell to kill its self. Towards this goal, we have performed preliminary experiments that test the impact of siRNA (RNA silencing) on diverse cellular signaling pathways, at unprecedented single-cell resolution. We propose ways in which a CAD system can process such data to automatically derive drug targets.

In the IC industry, chip design cycles are becoming more compressed, while designs themselves are growing in complexity. These trends necessitate efficient methods to handle late-stage engineering change orders (ECOs) to the functional specification, often in response to errors discovered after much of the implementation is finished. Past ECO synthesis algorithms have typically treated ECOs as functional errors and applied error diagnosis techniques to solve them. However, error diagnosis methods are primarily geared towards finding a single change, and moreover, tend to be computationally complex. In this paper, we propose a unique methodology that can systematically incorporate human intuition into the ECO process. Our methodology involves finding a set of directly substitutable points known as functional correspondences between the original implementation and the new specification by using name-preserving synthesis and user hints, to diminish the size of the ECO problem. On average, our approach can reduce the size of logic changes by 94% from those reported in current literature. We then incorporate our logic ECO changes into an incremental physical synthesis flow to demonstrate its usability in an industrial setting. Our ECO synthesis methodology is evaluated on high-performance industrial designs. Results indicate that post-ECO worst negative slack (WNS) improved 14% and total negative slack (TNS) improved 46% over pre-ECO.

Modern synthesis flows apply a series of technology independent optimization steps followed by mapping algorithms which bind the optimized network to a specific technology library. As the exact solution of the mapping problem is computationally intractable, algorithms used in practice use heuristic, typically tree-based approaches. The application of these algorithms results in mapped but suboptimal networks. In this work, we present a novel, efficient, and effective optimization algorithm for mapped networks which can be considered a generalization of SAT-sweeping. Our algorithm searches for alternative, more efficient implementations of each net in the network. Candidate support nets for reimplementation are selected using simulation signatures and verified using Boolean satisfiability. We report experimental results on the quality of our algorithm obtained from an implementation of the approach using the logic synthesis system ABC.

With the growing complexity of synthetic biological circuits, robust and systematic methods are needed for design and test. Leveraging lessons learned from the semiconductor and design automation industries, synthetic biologists are starting to adopt computer-aided design and verification software with some success. However, due to the great challenges associated with designing synthetic biological circuits, this nascent approach has to address many problems not present in electronic circuits. In this session, three leading synthetic biologists will share how they have developed software tools to help design and verify their synthetic circuits, the unique challenges they face, and their insights into the next generation of tools for synthetic biology.

The impact of physical synthesis on design performance is increasing as process technology scales. Current physical synthesis flows generally perform a series of individual netlist transformations based on local timing conditions. However, such optimizations lack sufficient perspective or scope to achieve timing closure in many cases. To address these issues, we develop an integrated transformation system that performs multiple optimizations simultaneously on larger design partitions than existing approaches. Our system, SPIRE, combines physically-aware register retiming, along with a novel form of cloning and register placement. SPIRE also incorporates a placement-dependent static timing analyzer (STA) with a delay model that accounts for buffering and is suitable for physical synthesis. Empirical results on 45 nm microprocessor designs show 8% improvement in worst-case slack and 69% improvement in total negative slack after an industrial physical synthesis flow was already completed.

State elements are increasingly vulnerable to soft errors due to their decreasing size, and the fact that latched errors cannot be completely eliminated by electrical or timing masking. Most prior methods of reducing the soft-error rate (SER) involve combinational redesign, which tends to add area and decrease testability, the latter a concern due to the prevalence of manufacturing defects. Our work explores the fundamental relations between the SER of sequential circuits and their testability in scan mode, and appears to be the first to improve both through retiming. Our retiming methodology relocates registers so that 1) registers become less observable with respect to primary outputs, thereby decreasing overall SER, and 2) combinational nodes become more observable with respect to registers (but not with respect to primary outputs), thereby increasing scan-testability. We present experimental results which show an average decrease of 42% in the SER of latches, and an average improvement of 31% random-pattern testability.

During the IC design process, functional specifications are often modified late in the design cycle, after placement and routing are completed. However, designers are left either to manually process such modifications by hand or to restart the design process from scratch---a very costly option. In order to address this issue, we present DeltaSyn, a method for generating a highly optimized logic difference between a modified high-level specification and an implemented design. DeltaSyn has the ability to locate boundaries in implemented logic within which changes can be confined. DeltaSyn demarcates the boundary in two phases. The first phase employs fast functional and structural analysis techniques to identify equivalent signals forming the input-side boundary of the changes. The second phase locates the outputside boundary of the changes through a novel dynamic algorithm that detects matching logic downstream from the changes required by the ECO. Experiments on industrial designs show that together these techniques successfully implement ECOs while preserving an average of 97% of the existing logic. Unlike previous approaches, the use of bitparallel logic simulation and fast SAT solvers enables high performance and scalability. DeltaSyn can process and verify a typical ECO for a design of around 10K gates in about 200 seconds or less.

We explore the use of signatures, i.e., partial truth tables generated via bit-parallel functional simulation, during soft error analysis and logic synthesis. We first present a signature-based CAD framework that incorporates tools for the logic-level Analysis of Soft Error Rate (x) and for Signature-based Design for Reliability (SiDeR). We observe that the soft error rate (SER) of a logic circuit is closely related to various testability parameters, such as signal observability and probability. We show that these parameters can be computed very efficiently (in linear time) by means of signatures. Consequently, AnSER evaluates logic masking two to three orders of magnitude faster than other SER evaluators while maintaining accuracy. AnSER can also compute SER efficiently in sequential circuits by approximating steady-state probabilities and sequential signal observabilities. In the second part of this paper, we incorporate AnSER into logic synthesis design flows aimed at reliable circuit design. SiDeR identifies and exploits redundancy already present in a circuit via signature comparison to decrease SER. We show that SiDeR reduces SER by 40% with only 13% area overhead. We also describe a second signature-based synthesis strategy that employs local rewriting to simultaneously improve area and decrease SER. This technique yields 13% reduction in SER with a 2% area decrease. We show that combining the two synthesis approaches can result in further area-reliability improvements.

Soft errors, once only of concern in memories, are beginning to affect logic as well. Determining the soft error rate (SER) of a combinational circuit involves three main masking mechanisms: logic, timing and electrical. Most previous papers focus on logic and electrical masking. In this paper we develop static and statistical analysis techniques for timing masking that estimate the error-latching window of each gate. Our SER evaluation algorithms incorporating timing masking are orders of magnitude faster than comparable evaluators and can be used in synthesis and layout. We show that 62 % of gates identified as error-critical using timing masking would not be identifiable by considering only logic masking. Furthermore, hardening the top 10 % of errorcritical gates leads to a 43 % reduction in the SER. We also propose a more subtle solution, gate-relocation for technologies where wire delay dominates gate delay. We decrease the error-latching window of each gate by relocating it in such a way that path lengths to primary outputs are equalized. Our results show a 14 % improvement in SER with no area overhead. 1

We propose the probabilistic transfer matrix (PTM) framework to capture nondeterministic behavior in logic circuits. PTMs provide a concise description of both normal and faulty behavior, and are well-suited to reliability and error susceptibility calculations. A few simple composition rules based on connectivity can be used to recursively build larger PTMs (representing entire logic circuits) from smaller gate PTMs. PTMs for gates in series are combined using matrix multiplication, and PTMs for gates in parallel are combined using the tensor product operation. PTMs can accurately calculate joint output probabilities in the presence of reconvergent fanout and inseparable joint input distributions. To improve computational efficiency, we encode PTMs as algebraic decision diagrams (ADDs). We also develop equivalent ADD algorithms for newly defined matrix operations such as eliminate_variables and eliminate_redundant_variables , which aid in the numerical computation of circuit PTMs. We use PTMs to evaluate circuit reliability and derive polynomial approximations for circuit error probabilities in terms of gate error probabilities. PTMs can also analyze the effects of logic and electrical masking on error mitigation. We show that ignoring logic masking can overestimate errors by an order of magnitude. We incorporate electrical masking by computing error attenuation probabilities, based on analytical models, into an extended PTM framework for reliability computation. We further define a susceptibility measure to identify gates whose errors are not well masked. We show that hardening a few gates can significantly improve circuit reliability.

While circuit density and power efficiency increase with each major advance in IC technology, reliability with respect to soft errors tends to decrease. Current solutions to this problem such as TMR require high area and power overhead. In this work, soft-error reliability is improved with minimal area overhead by careful, localized circuit restructuring. The key idea is to increase logic masking of errors by taking advantage of conditions already present in the circuit, such as observability don't-cares. We describe two circuit modification techniques to improve reliability: don't-care-based resynthesis and local rewriting. A key feature of these techniques is fast, on-the-fly estimation of soft error rate (SER) using our reliability evaluator AnSER. This tool is compared against prior SER evaluators and found to run orders of magnitude faster. We show empirically that our reliability-driven synthesis methods can reduce SER by 29-40% with only 5-13% area overhead.

Soft errors are an increasingly serious problem for logic circuits. To estimate the effects of soft errors on such circuits, we develop a general computational framework based on probabilistic transfer matrices (PTMs). In particular, we apply them to evaluate circuit reliability in the presence of soft errors, which involves combining the PTMs of gates to form an overall circuit PTM. Information, such as output probabilities, the overall probability of error, and signal observability, can then be extracted from the circuit PTM. We employ algebraic decision diagrams (ADDs) to improve the efficiency of PTM operations. A particularly challenging technical problem, solved in our work, is to extend simultaneously tensor products and matrix multiplication in terms of ADDs to non-square matrices. Our PTM-based method enables accurate evaluation of reliability for moderately large circuits and can be extended by circuit partitioning. To demonstrate the power of the PTM approach, we apply it to several problems in fault-tolerant design and reliability improvement.

Transient faults are becoming an increasingly serious concern for logic circuits. They can be caused by thermal neutrons, present at all altitudes, and by other types of ionizing radiation, especially in aerospace applications and nuclear engineering. In this paper we examine issues related to detection of transient errors. The difficulty in testing for transient errors is that they are not always present. Test vectors need to be repeated a number of times in order to detect a fault. We show how to compute a measure for the detectability of transient faults with respect to specific test vectors. This is done using a matrix-based gate-fault model known as the probabilistic transfer matrix model. Using this detectability measure we derive methods to generate multisets of tests to verify probability distributions of faults and detect abnormalities in circuit behavior. Applications of this method include detection of increased atmospheric radiation in terms of its impact on circuits, and testing for process variation that increases the susceptibility of a circuit to transient errors.