Author name cluster

Michael J. Black

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

12 papers

2 author rows

ICLR Conference 2025 Conference Paper

Can Large Language Models Understand Symbolic Graphics Programs?

Zeju Qiu
Weiyang Liu
Haiwen Feng
Zhen Liu 0019
Tim Z. Xiao
Katherine M. Collins
Joshua B. Tenenbaum
Adrian Weller

Against the backdrop of enthusiasm for large language models (LLMs), there is a growing need to scientifically assess their capabilities and shortcomings. This is nontrivial in part because it is difficult to find tasks which the models have not encountered during training. Utilizing symbolic graphics programs, we propose a domain well-suited to test multiple spatial-semantic reasoning skills of LLMs. Popular in computer graphics, these programs procedurally generate visual data. While LLMs exhibit impressive skills in general program synthesis and analysis, symbolic graphics programs offer a new layer of evaluation: they allow us to test an LLM's ability to answer semantic questions about the images or 3D geometries without a vision encoder. To semantically understand the symbolic programs, LLMs would need to possess the ability to "imagine" and reason how the corresponding graphics content would look with only the symbolic description of the local curvatures and strokes. We use this task to evaluate LLMs by creating a large benchmark for the semantic visual understanding of symbolic graphics programs, built procedurally with minimal human effort. Particular emphasis is placed on transformations of images that leave the image level semantics invariant while introducing significant changes to the underlying program. We evaluate commercial and open-source LLMs on our benchmark to assess their ability to reason about visual output of programs, finding that LLMs considered stronger at reasoning generally perform better. Lastly, we introduce a novel method to improve this ability -- Symbolic Instruction Tuning (SIT), in which the LLM is finetuned with pre-collected instruction data on symbolic graphics programs. Interestingly, we find that SIT not only improves LLM's understanding on symbolic programs, but it also improves general reasoning ability on various other benchmarks.

Details

ICLR Conference 2024 Conference Paper

Ghost on the Shell: An Expressive Representation of General 3D Shapes

Zhen Liu 0019
Yao Feng 0001
Yuliang Xiu
Weiyang Liu
Liam Paull
Michael J. Black
Bernhard Schölkopf

The creation of photorealistic virtual worlds requires the accurate modeling of 3D surface geometry for a wide range of objects. For this, meshes are appealing since they enable 1) fast physics-based rendering with realistic material and lighting, 2) physical simulation, and 3) are memory-efficient for modern graphics pipelines. Recent work on reconstructing and statistically modeling 3D shape, however, has critiqued meshes as being topologically inflexible. To capture a wide range of object shapes, any 3D representation must be able to model solid, watertight, shapes as well as thin, open, surfaces. Recent work has focused on the former, and methods for reconstructing open surfaces do not support fast reconstruction with material and lighting or unconditional generative modelling. Inspired by the observation that open surfaces can be seen as islands floating on watertight surfaces, we parametrize open surfaces by defining a manifold signed distance field on watertight templates. With this parametrization, we further develop a grid-based and differentiable representation that parametrizes both watertight and non-watertight meshes of arbitrary topology. Our new representation, called Ghost-on-the-Shell (G-Shell), enables two important applications: differentiable rasterization-based reconstruction from multiview images and generative modelling of non-watertight meshes. We empirically demonstrate that G-Shell achieves state-of-the-art performance on non-watertight mesh reconstruction and generation tasks, while also performing effectively for watertight meshes.

Details

ICLR Conference 2024 Conference Paper

Parameter-Efficient Orthogonal Finetuning via Butterfly Factorization

Weiyang Liu
Zeju Qiu
Yao Feng 0001
Yuliang Xiu
Yuxuan Xue 0001
Longhui Yu
Haiwen Feng
Zhen Liu 0019

Large foundation models are becoming ubiquitous, but training them from scratch is prohibitively expensive. Thus, efficiently adapting these powerful models to downstream tasks is increasingly important. In this paper, we study a principled finetuning paradigm -- Orthogonal Finetuning (OFT) -- for downstream task adaptation. Despite demonstrating good generalizability, OFT still uses a fairly large number of trainable parameters due to the high dimensionality of orthogonal matrices. To address this, we start by examining OFT from an information transmission perspective, and then identify a few key desiderata that enable better parameter-efficiency. Inspired by how the Cooley-Tukey fast Fourier transform algorithm enables efficient information transmission, we propose an efficient orthogonal parameterization using butterfly structures. We apply this parameterization to OFT, creating a novel parameter-efficient finetuning method, called Orthogonal Butterfly (BOFT). By subsuming OFT as a special case, BOFT introduces a generalized orthogonal finetuning framework. Finally, we conduct an extensive empirical study of adapting large vision transformers, large language models, and text-to-image diffusion models to various downstream tasks in computer vision and natural language. The results validate the effectiveness of BOFT as a generic finetuning method.

Details

TMLR Journal 2024 Journal Article

Re-Thinking Inverse Graphics With Large Language Models

Peter Kulits
Haiwen Feng
Weiyang Liu
Victoria Fernandez Abrevaya
Michael J. Black

Inverse graphics -- the task of inverting an image into physical variables that, when rendered, enable reproduction of the observed scene -- is a fundamental challenge in computer vision and graphics. Successfully disentangling an image into its constituent elements, such as the shape, color, and material properties of the objects of the 3D scene that produced it, requires a comprehensive understanding of the environment. This complexity limits the ability of existing carefully engineered approaches to generalize across domains. Inspired by the zero-shot ability of large language models (LLMs) to generalize to novel contexts, we investigate the possibility of leveraging the broad world knowledge encoded in such models to solve inverse-graphics problems. To this end, we propose the Inverse-Graphics Large Language Model (IG-LLM), an inverse-graphics framework centered around an LLM, that autoregressively decodes a visual embedding into a structured, compositional 3D-scene representation. We incorporate a frozen pre-trained visual encoder and a continuous numeric head to enable end-to-end training. Through our investigation, we demonstrate the potential of LLMs to facilitate inverse graphics through next-token prediction, without the application of image-space supervision. Our analysis enables new possibilities for precise spatial reasoning about images that exploit the visual knowledge of LLMs. We release our code and data at https://ig-llm.is.tue.mpg.de/ to ensure the reproducibility of our investigation and to facilitate future research.

PDF Details

ICLR Conference 2023 Conference Paper

MeshDiffusion: Score-based Generative 3D Mesh Modeling

Zhen Liu 0019
Yao Feng 0001
Michael J. Black
Derek Nowrouzezahrai
Liam Paull
Weiyang Liu

We consider the task of generating realistic 3D shapes, which is useful for a variety of applications such as automatic scene generation and physical simulation. Compared to other 3D representations like voxels and point clouds, meshes are more desirable in practice, because (1) they enable easy and arbitrary manipulation of shapes for relighting and simulation, and (2) they can fully leverage the power of modern graphics pipelines which are mostly optimized for meshes. Previous scalable methods for generating meshes typically rely on sub-optimal post-processing, and they tend to produce overly-smooth or noisy surfaces without fine-grained geometric details. To overcome these shortcomings, we take advantage of the graph structure of meshes and use a simple yet very effective generative modeling method to generate 3D meshes. Specifically, we represent meshes with deformable tetrahedral grids, and then train a diffusion model on this direct parameterization. We demonstrate the effectiveness of our model on multiple generative tasks.

Details

IROS Conference 2022 Conference Paper

Deep Residual Reinforcement Learning based Autonomous Blimp Control

Yu Tang Liu
Eric Price 0002
Michael J. Black
Aamir Ahmad

Blimps are well suited to perform long-duration aerial tasks as they are energy efficient, relatively silent and safe. To address the blimp navigation and control task, in previous work we developed a hardware and software-in-the-loop framework and a PID-based controller for large blimps in the presence of wind disturbance. However, blimps have a deformable structure and their dynamics are inherently non-linear and time-delayed, making PID controllers difficult to tune. Thus, often resulting in large tracking errors. Moreover, the buoyancy of a blimp is constantly changing due to variations in ambient temperature and pressure. To address these issues, in this paper we present a learning-based framework based on deep residual reinforcement learning (DRRL), for the blimp control task. Within this framework, we first employ a PID controller to provide baseline performance. Subsequently, the DRRL agent learns to modify the PID decisions by interaction with the environment. We demonstrate in simulation that DRRL agent consistently improves the PID performance. Through rigorous simulation experiments, we show that the agent is robust to changes in wind speed and buoyancy. In real-world experiments, we demonstrate that the agent, trained only in simulation, is sufficiently robust to control an actual blimp in windy conditions. We openly provide the source code of our approach at https://github.com/robot-perception-group/AutonomousBlimpDRL.Video demonstration is provided at https://youtu.be/EMC4KnlH0yI.

Details

YNIMG Journal 2022 Journal Article

The neural coding of face and body orientation in occipitotemporal cortex

Celia Foster
Mintao Zhao
Timo Bolkart
Michael J. Black
Andreas Bartels
Isabelle Bülthoff

Face and body orientation convey important information for us to understand other people's actions, intentions and social interactions. It has been shown that several occipitotemporal areas respond differently to faces or bodies of different orientations. However, whether face and body orientation are processed by partially overlapping or completely separate brain networks remains unclear, as the neural coding of face and body orientation is often investigated separately. Here, we recorded participants' brain activity using fMRI while they viewed faces and bodies shown from three different orientations, while attending to either orientation or identity information. Using multivoxel pattern analysis we investigated which brain regions process face and body orientation respectively, and which regions encode both face and body orientation in a stimulus-independent manner. We found that patterns of neural responses evoked by different stimulus orientations in the occipital face area, extrastriate body area, lateral occipital complex and right early visual cortex could generalise across faces and bodies, suggesting a stimulus-independent encoding of person orientation in occipitotemporal cortex. This finding was consistent across functionally defined regions of interest and a whole-brain searchlight approach. The fusiform face area responded to face but not body orientation, suggesting that orientation responses in this area are face-specific. Moreover, neural responses to orientation were remarkably consistent regardless of whether participants attended to the orientation of faces and bodies or not. Together, these results demonstrate that face and body orientation are processed in a partially overlapping brain network, with a stimulus-independent neural code for face and body orientation in occipitotemporal cortex.

Details DOI

ICLR Conference 2020 Conference Paper

From Variational to Deterministic Autoencoders

Partha Ghosh
Mehdi S. M. Sajjadi
Antonio Vergari
Michael J. Black
Bernhard Schölkopf

Variational Autoencoders (VAEs) provide a theoretically-backed and popular framework for deep generative models. However, learning a VAE from data poses still unanswered theoretical questions and considerable practical challenges. In this work, we propose an alternative framework for generative modeling that is simpler, easier to train, and deterministic, yet has many of the advantages of the VAE. We observe that sampling a stochastic encoder in a Gaussian VAE can be interpreted as simply injecting noise into the input of a deterministic decoder. We investigate how substituting this kind of stochasticity, with other explicit and implicit regularization schemes, can lead to an equally smooth and meaningful latent space without having to force it to conform to an arbitrarily chosen prior. To retrieve a generative mechanism to sample new data points, we introduce an ex-post density estimation step that can be readily applied to the proposed framework as well as existing VAEs, improving their sample quality. We show, in a rigorous empirical study, that the proposed regularized deterministic autoencoders are able to generate samples that are comparable to, or better than, those of VAEs and more powerful alternatives when applied to images as well as to structured data such as molecules.

Details

YNIMG Journal 2019 Journal Article

Decoding subcategories of human bodies from both body- and face-responsive cortical regions

Celia Foster
Mintao Zhao
Javier Romero
Michael J. Black
Betty J. Mohler
Andreas Bartels
Isabelle Bülthoff

Our visual system can easily categorize objects (e. g. faces vs. bodies) and further differentiate them into subcategories (e. g. male vs. female). This ability is particularly important for objects of social significance, such as human faces and bodies. While many studies have demonstrated category selectivity to faces and bodies in the brain, how subcategories of faces and bodies are represented remains unclear. Here, we investigated how the brain encodes two prominent subcategories shared by both faces and bodies, sex and weight, and whether neural responses to these subcategories rely on low-level visual, high-level visual or semantic similarity. We recorded brain activity with fMRI while participants viewed faces and bodies that varied in sex, weight, and image size. The results showed that the sex of bodies can be decoded from both body- and face-responsive brain areas, with the former exhibiting more consistent size-invariant decoding than the latter. Body weight could also be decoded in face-responsive areas and in distributed body-responsive areas, and this decoding was also invariant to image size. The weight of faces could be decoded from the fusiform body area (FBA), and weight could be decoded across face and body stimuli in the extrastriate body area (EBA) and a distributed body-responsive area. The sex of well-controlled faces (e. g. excluding hairstyles) could not be decoded from face- or body-responsive regions. These results demonstrate that both face- and body-responsive brain regions encode information that can distinguish the sex and weight of bodies. Moreover, the neural patterns corresponding to sex and weight were invariant to image size and could sometimes generalize across face and body stimuli, suggesting that such subcategorical information is encoded with a high-level visual or semantic code.

Details DOI

AAAI Conference 2019 Conference Paper

Resisting Adversarial Attacks Using Gaussian Mixture Variational Autoencoders

Partha Ghosh
Arpan Losalka
Michael J. Black

Susceptibility of deep neural networks to adversarial attacks poses a major theoretical and practical challenge. All efforts to harden classifiers against such attacks have seen limited success till now. Two distinct categories of samples against which deep neural networks are vulnerable, “adversarial samples” and “fooling samples”, have been tackled separately so far due to the difficulty posed when considered together. In this work, we show how one can defend against them both under a unified framework. Our model has the form of a variational autoencoder with a Gaussian mixture prior on the latent variable, such that each mixture component corresponds to a single class. We show how selective classification can be performed using this model, thereby causing the adversarial objective to entail a conflict. The proposed method leads to the rejection of adversarial samples instead of misclassification, while maintaining high precision and recall on test data. It also inherently provides a way of learning a selective classifier in a semi-supervised scenario, which can similarly resist adversarial attacks. We further show how one can reclassify the detected adversarial samples by iterative optimization. 1

PDF Details

ICML Conference 2014 Conference Paper

Preserving Modes and Messages via Diverse Particle Selection

Jason Pacheco
Silvia Zuffi
Michael J. Black
Erik B. Sudderth

In applications of graphical models arising in domains such as computer vision and signal processing, we often seek the most likely configurations of high-dimensional, continuous variables. We develop a particle-based max-product algorithm which maintains a diverse set of posterior mode hypotheses, and is robust to initialization. At each iteration, the set of hypotheses at each node is augmented via stochastic proposals, and then reduced via an efficient selection algorithm. The integer program underlying our optimization-based particle selection minimizes errors in subsequent max-product message updates. This objective automatically encourages diversity in the maintained hypotheses, without requiring tuning of application-specific distances among hypotheses. By avoiding the stochastic resampling steps underlying particle sum-product algorithms, we also avoid common degeneracies where particles collapse onto a single hypothesis. Our approach significantly outperforms previous particle-based algorithms in experiments focusing on the estimation of human pose from single images.

Details

AAAI Conference 1990 Conference Paper

Constraints for the Early Detection of Discontinuity from Motion

Michael J. Black
P. Anandan

Surface discontinuities are detected in a sequence of images by exploiting physical constraints at early stages in the processing of visual motion. To achieve accurate early discontinuity detection we exploit five physical constraints on the presence of discontinuities: i) the shape of the sum of squared differences (SSD) error surface in the presence of surface discontinuities; G) the change in the shape of the SSD surface due to relative surface motion; G) distribution of optic flow in a neighborhood of a discontinuity; iv) spatial consistency of discontinuities; V) temporal consistency of discontinuities. The constraints are described, and experimental results on sequences of real and synthetic images are presented. The work has applications in the recovery of environmental structure from motion and in the generation of dense optic flow fields.

PDF Details