Kartic Subr

Sean Memery, Osmar Cedron, Kartic Subr

Artistic authoring of 3D environments is a laborious enterprise that also requires skilled content creators. There have been impressive improvements in using machine learning to address different aspects of generating 3D content, such as generating meshes, arranging geometry, synthesizing textures, etc. In this paper we develop a model to generate Bidirectional Reflectance Distribution Functions (BRDFs) from descriptive textual prompts. BRDFs are four dimensional probability distributions that characterize the interaction of light with surface materials. They are either represented parametrically, or by tabulating the probability density associated with every pair of incident and outgoing angles. The former lends itself to artistic editing while the latter is used when measuring the appearance of real materials. Numerous works have focused on hypothesizing BRDF models from images of materials. We learn a mapping from textual descriptions of materials to parametric BRDFs. Our model is first trained using a semi-supervised approach before being tuned via an unsupervised scheme. Although our model is general, in this paper we specifically generate parameters for MDL materials, conditioned on natural language descriptions, within NVIDIA's Omniverse platform. This enables use cases such as real-time text prompts to change materials of objects in 3D environments such as "dull plastic" or "shiny iron". Since the output of our model is a parametric BRDF, rather than an image of the material, it may be used to render materials using any shape under arbitrarily specified viewing and lighting conditions.

Sean Memery, Kartic Subr

Artificial intelligence (AI) agents embedded in environments with physics-based interaction face many challenges including reasoning, planning, summarization, and question answering. This problem is exacerbated when a human user wishes to either guide or interact with the agent in natural language. Although the use of Language Models (LMs) is the default choice, as an AI tool, they struggle with tasks involving physics. The LM's capability for physical reasoning is learned from observational data, rather than being grounded in simulation. A common approach is to include simulation traces as context, but this suffers from poor scalability as simulation traces contain larger volumes of fine-grained numerical and semantic data. In this paper, we propose a natural language guided method to discover coarse-grained patterns (e.g., 'rigid-body collision', 'stable support', etc.) from detailed simulation logs. Specifically, we synthesize programs that operate on simulation logs and map them to a series of high level activated patterns. We show, through two physics benchmarks, that this annotated representation of the simulation log is more amenable to natural language reasoning about physical systems. We demonstrate how this method enables LMs to generate effective reward programs from goals specified in natural language, which may be used within the context of planning or supervised learning.

Kevin Yandoka Denamganaï, Kartic Subr

Despite emerging research on Language Models (LM), few approaches analyse the invertibility of LMs. That is, given a LM and a desirable target output sequence of tokens, determining what input prompts would yield the target output remains an open problem. We formulate this problem as a classical gradient-based optimisation. First, we propose a simple algorithm to achieve end-to-end differentiability of a given (frozen) LM and then find optimised prompts via gradient descent. Our central insight is to view LMs as functions operating on sequences of distributions over tokens (rather than the traditional view as functions on sequences of tokens). Our experiments and ablations demonstrate that our DLM-powered inversion can reliably and efficiently optimise prompts of lengths $10$ and $80$ for targets of length $20$, for several white-box LMs (out-of-the-box).

Sean Memery, Mirella Lapata, Kartic Subr

Several machine learning methods aim to learn or reason about complex physical systems. A common first-step towards reasoning is to infer system parameters from observations of its behavior. In this paper, we investigate the performance of Large Language Models (LLMs) at performing parameter inference in the context of physical systems. Our experiments suggest that they are not inherently suited to this task, even for simple systems. We propose a promising direction of exploration, which involves the use of physical simulators to augment the context of LLMs. We assess and compare the performance of different LLMs on a simple example with and without access to physical simulation.

Sean Memery, Kevin Denamganai, Jiaxin Zhang et al.

We present an interactive and explainable automated coaching assistant called CueTip for a variant of pool/billiards. CueTip's novelty lies in its combination of three features: a natural-language interface, an ability to perform contextual, physics-aware reasoning, and that its explanations are rooted in a set of predetermined guidelines developed by domain experts. We instrument a physics simulator so that it generates event traces in natural language alongside traditional state traces. Event traces lend themselves to interpretation by language models, which serve as the interface to our assistant. We design and train a neural adaptor that decouples tactical choices made by CueTip from its interactivity and explainability allowing it to be reconfigured to mimic any pool playing agent. Our experiments show that CueTip enables contextual query-based assistance and explanations while maintaining the strength of the agent in terms of win rate (improving it in some situations). The explanations generated by CueTip are physically-aware and grounded in the expert rules and are therefore more reliable.

Jianning Deng, Kartic Subr, Hakan Bilen

We present a novel self-supervised framework for learning articulated object representations from sparse-view, unposed images. Unlike prior methods that require dense multi-view observations and ground-truth camera poses, our approach operates with as few as four views per articulation and no camera supervision. To address the inherent challenges, we first reconstruct each articulation independently using recent advances in sparse-view 3D reconstruction, then learn a deformation field that establishes dense correspondences across poses. A progressive disentanglement strategy further separates static from moving parts, enabling robust separation of camera and object motion. Finally, we jointly optimize geometry, appearance, and kinematics with a self-supervised loss that enforces cross-view and cross-pose consistency. Experiments on the standard benchmark and real-world examples demonstrate that our method produces accurate and detailed articulated object representations under significantly weaker input assumptions than existing approaches.

Sean Memery, Kevin Denamganai, Anna Kapron-King et al.

Inverse problems are central to a wide range of fields, including healthcare, climate science, and agriculture. They involve the estimation of inputs, typically via iterative optimization, to some known forward model so that it produces a desired outcome. Despite considerable development in the explainability and interpretability of forward models, the iterative optimization of inverse problems remains largely cryptic to domain experts. We propose a methodology to produce explanations, from traces produced by an optimizer, that are interpretable by humans at the abstraction of the domain. The central idea in our approach is to instrument a differentiable simulator so that it emits natural language events during its forward and backward passes. In a post-process, we use a Language Model to create an explanation from the list of events. We demonstrate the effectiveness of our approach with an illustrative optimization problem and an example involving the training of a neural network.

Jianning Deng, Kartic Subr, Hakan Bilen

We propose a novel unsupervised method to learn the pose and part-segmentation of articulated objects with rigid parts. Given two observations of an object in different articulation states, our method learns the geometry and appearance of object parts by using an implicit model from the first observation, distils the part segmentation and articulation from the second observation while rendering the latter observation. Additionally, to tackle the complexities in the joint optimization of part segmentation and articulation, we propose a voxel grid-based initialization strategy and a decoupled optimization procedure. Compared to the prior unsupervised work, our model obtains significantly better performance, and generalizes to objects with multiple parts while it can be efficiently from few views for the latter observation.

Alexandros Dimitrios Keros, Vidit Nanda, Kartic Subr

Simplicial complexes can be viewed as high dimensional generalizations of graphs that explicitly encode multi-way ordered relations between vertices at different resolutions, all at once. This concept is central towards detection of higher dimensional topological features of data, features to which graphs, encoding only pairwise relationships, remain oblivious. While attempts have been made to extend Graph Neural Networks (GNNs) to a simplicial complex setting, the methods do not inherently exploit, or reason about, the underlying topological structure of the network. We propose a graph convolutional model for learning functions parametrized by the $k$-homological features of simplicial complexes. By spectrally manipulating their combinatorial $k$-dimensional Hodge Laplacians, the proposed model enables learning topological features of the underlying simplicial complexes, specifically, the distance of each $k$-simplex from the nearest "optimal" $k$-th homology generator, effectively providing an alternative to homology localization.

Leonardo V. Castorina, Rokas Petrenas, Kartic Subr et al.

Proteins perform critical processes in all living systems: converting solar energy into chemical energy, replicating DNA, as the basis of highly performant materials, sensing and much more. While an incredible range of functionality has been sampled in nature, it accounts for a tiny fraction of the possible protein universe. If we could tap into this pool of unexplored protein structures, we could search for novel proteins with useful properties that we could apply to tackle the environmental and medical challenges facing humanity. This is the purpose of protein design. Sequence design is an important aspect of protein design, and many successful methods to do this have been developed. Recently, deep-learning methods that frame it as a classification problem have emerged as a powerful approach. Beyond their reported improvement in performance, their primary advantage over physics-based methods is that the computational burden is shifted from the user to the developers, thereby increasing accessibility to the design method. Despite this trend, the tools for assessment and comparison of such models remain quite generic. The goal of this paper is to both address the timely problem of evaluation and to shine a spotlight, within the Machine Learning community, on specific assessment criteria that will accelerate impact. We present a carefully curated benchmark set of proteins and propose a number of standard tests to assess the performance of deep learning based methods. Our robust benchmark provides biological insight into the behaviour of design methods, which is essential for evaluating their performance and utility. We compare five existing models with two novel models for sequence prediction. Finally, we test the designs produced by these models with AlphaFold2, a state-of-the-art structure-prediction algorithm, to determine if they are likely to fold into the intended 3D shapes.

Tatiana Lopez-Guevara, Michael Burke, Nicholas K. Taylor et al.

Model-free reinforcement learning (RL) is a powerful tool to learn a broad range of robot skills and policies. However, a lack of policy interpretability can inhibit their successful deployment in downstream applications, particularly when differences in environmental conditions may result in unpredictable behaviour or generalisation failures. As a result, there has been a growing emphasis in machine learning around the inclusion of stronger inductive biases in models to improve generalisation. This paper proposes an alternative strategy, inverse value estimation for interpretable policy certificates (IV-Posterior), which seeks to identify the inductive biases or idealised conditions of operation already held by pre-trained policies, and then use this information to guide their deployment. IV-Posterior uses MaskedAutoregressive Flows to fit distributions over the set of conditions or environmental parameters in which a policy is likely to be effective. This distribution can then be used as a policy certificate in downstream applications. We illustrate the use of IV-Posterior across a two environments, and show that substantial performance gains can be obtained when policy selection incorporates knowledge of the inductive biases that these policies hold.

Michael Burke, Kartic Subr, Subramanian Ramamoorthy

Humans can easily reason about the sequence of high level actions needed to complete tasks, but it is particularly difficult to instil this ability in robots trained from relatively few examples. This work considers the task of neural action sequencing conditioned on a single reference visual state. This task is extremely challenging as it is not only subject to the significant combinatorial complexity that arises from large action sets, but also requires a model that can perform some form of symbol grounding, mapping high dimensional input data to actions, while reasoning about action relationships. This paper takes a permutation perspective and argues that action sequencing benefits from the ability to reason about both permutations and ordering concepts. Empirical analysis shows that neural models trained with latent permutations outperform standard neural architectures in constrained action sequencing tasks. Results also show that action sequencing using visual permutations is an effective mechanism to initialise and speed up traditional planning techniques and successfully scales to far greater action set sizes than models considered previously.

Kartic Subr

Many applications require the calculation of integrals of multidimensional functions. A general and popular procedure is to estimate integrals by averaging multiple evaluations of the function. Often, each evaluation of the function entails costly computations. The use of a \emph{proxy} or surrogate for the true function is useful if repeated evaluations are necessary. The proxy is even more useful if its integral is known analytically and can be calculated practically. We propose the use of a versatile yet simple class of artificial neural networks -- sigmoidal universal approximators -- as a proxy for functions whose integrals need to be estimated. We design a family of fixed networks, which we call Q-NETs, that operate on parameters of a trained proxy to calculate exact integrals over \emph{any subset of dimensions} of the input domain. We identify transformations to the input space for which integrals may be recalculated without resampling the integrand or retraining the proxy. We highlight the benefits of this scheme for a few applications such as inverse rendering, generation of procedural noise, visualization and simulation. The proposed proxy is appealing in the following contexts: the dimensionality is low ($<10$D); the estimation of integrals needs to be decoupled from the sampling strategy; sparse, adaptive sampling is used; marginal functions need to be known in functional form; or when powerful Single Instruction Multiple Data/Thread (SIMD/SIMT) pipelines are available for computation.

Rita Pucci, Jitendra Shankaraiah, Devcharan Jathanna et al.

Photographs of wild animals in their natural habitats can be recorded unobtrusively via cameras that are triggered by motion nearby. The installation of such camera traps is becoming increasingly common across the world. Although this is a convenient source of invaluable data for biologists, ecologists and conservationists, the arduous task of poring through potentially millions of pictures each season introduces prohibitive costs and frustrating delays. We develop automatic algorithms that are able to detect animals, identify the species of animals and to recognize individual animals for two species. we propose the first fully-automatic tool that can recognize specific individuals of leopard and tiger due to their characteristic body markings. We adopt a class of supervised learning approach of machine learning where a Deep Convolutional Neural Network (DCNN) is trained using several instances of manually-labelled images for each of the three classification tasks. We demonstrate the effectiveness of our approach on a data set of camera-trap images recorded in the jungles of Southern India.

Michael Burke, Katie Lu, Daniel Angelov et al.

Informative path-planning is a well established approach to visual-servoing and active viewpoint selection in robotics, but typically assumes that a suitable cost function or goal state is known. This work considers the inverse problem, where the goal of the task is unknown, and a reward function needs to be inferred from exploratory example demonstrations provided by a demonstrator, for use in a downstream informative path-planning policy. Unfortunately, many existing reward inference strategies are unsuited to this class of problems, due to the exploratory nature of the demonstrations. In this paper, we propose an alternative approach to cope with the class of problems where these sub-optimal, exploratory demonstrations occur. We hypothesise that, in tasks which require discovery, successive states of any demonstration are progressively more likely to be associated with a higher reward, and use this hypothesis to generate time-based binary comparison outcomes and infer reward functions that support these ranks, under a probabilistic generative model. We formalise this \emph{probabilistic temporal ranking} approach and show that it improves upon existing approaches to perform reward inference for autonomous ultrasound scanning, a novel application of learning from demonstration in medical imaging while also being of value across a broad range of goal-oriented learning from demonstration tasks. \keywords{Visual servoing \and reward inference \and probabilistic temporal ranking

Martin Asenov, Michael Burke, Daniel Angelov et al.

Videos provide a rich source of information, but it is generally hard to extract dynamical parameters of interest. Inferring those parameters from a video stream would be beneficial for physical reasoning. Robots performing tasks in dynamic environments would benefit greatly from understanding the underlying environment motion, in order to make future predictions and to synthesize effective control policies that use this inductive bias. Online physical reasoning is therefore a fundamental requirement for robust autonomous agents. When the dynamics involves multiple modes (due to contacts or interactions between objects) and sensing must proceed directly from a rich sensory stream such as video, then traditional methods for system identification may not be well suited. We propose an approach wherein fast parameter estimation can be achieved directly from video. We integrate a physically based dynamics model with a recurrent variational autoencoder, by introducing an additional loss to enforce desired constraints. The model, which we call Vid2Param, can be trained entirely in simulation, in an end-to-end manner with domain randomization, to perform online system identification, and make probabilistic forward predictions of parameters of interest. This enables the resulting model to encode parameters such as position, velocity, restitution, air drag and other physical properties of the system. We illustrate the utility of this in physical experiments wherein a PR2 robot with a velocity constrained arm must intercept an unknown bouncing ball with partly occluded vision, by estimating the physical parameters of this ball directly from the video trace after the ball is released.

Tatiana Lopez-Guevara, Rita Pucci, Nicholas Taylor et al.

Our brains are able to exploit coarse physical models of fluids to solve everyday manipulation tasks. There has been considerable interest in developing such a capability in robots so that they can autonomously manipulate fluids adapting to different conditions. In this paper, we investigate the problem of adaptation to liquids with different characteristics. We develop a simple calibration task (stirring with a stick) that enables rapid inference of the parameters of the liquid from RBG data. We perform the inference in the space of simulation parameters rather than on physically accurate parameters. This facilitates prediction and optimization tasks since the inferred parameters may be fed directly to the simulator. We demonstrate that our "stirring" learner performs better than when the robot is calibrated with pouring actions. We show that our method is able to infer properties of three different liquids -- water, glycerin and gel -- and present experimental results by executing stirring and pouring actions on a UR10. We believe that decoupling of the training actions from the goal task is an important step towards simple, autonomous learning of the behavior of different fluids in unstructured environments.

Martin Asenov, Marius Rutkauskas, Derryck Reid et al.

Sensors are routinely mounted on robots to acquire various forms of measurements in spatio-temporal fields. Locating features within these fields and reconstruction (mapping) of the dense fields can be challenging in resource-constrained situations, such as when trying to locate the source of a gas leak from a small number of measurements. In such cases, a model of the underlying complex dynamics can be exploited to discover informative paths within the field. We use a fluid simulator as a model, to guide inference for the location of a gas leak. We perform localization via minimization of the discrepancy between observed measurements and gas concentrations predicted by the simulator. Our method is able to account for dynamically varying parameters of wind flow (e.g., direction and strength), and its effects on the observed distribution of gas. We develop algorithms for off-line inference as well as for on-line path discovery via active sensing. We demonstrate the efficiency, accuracy and versatility of our algorithm using experiments with a physical robot conducted in outdoor environments. We deploy an unmanned air vehicle (UAV) mounted with a CO2 sensor to automatically seek out a gas cylinder emitting CO2 via a nozzle. We evaluate the accuracy of our algorithm by measuring the error in the inferred location of the nozzle, based on which we show that our proposed approach is competitive with respect to state of the art baselines.

Paul Henderson, Kartic Subr, Vittorio Ferrari

Efficient authoring of vast virtual environments hinges on algorithms that are able to automatically generate content while also being controllable. We propose a method to automatically generate furniture layouts for indoor environments. Our method is simple, efficient, human-interpretable and amenable to a wide variety of constraints. We model the composition of rooms into classes of objects and learn joint (co-occurrence) statistics from a database of training layouts. We generate new layouts by performing a sequence of conditional sampling steps, exploiting the statistics learned from the database. The generated layouts are specified as 3D object models, along with their positions and orientations. We show they are of equivalent perceived quality to the training layouts, and compare favorably to a state-of-the-art method. We incorporate constraints using a general mechanism -- rejection sampling -- which provides great flexibility at the cost of extra computation. We demonstrate the versatility of our method by applying a wide variety of constraints relevant to real-world applications.