If you ask a human to describe an image, they might do so in a thousand different ways. Traditionally, image captioning models are trained to approximate the reference distribution of image captions, however, doing so encourages captions that are viewpoint-impoverished. Such captions often focus on only a subset of the possible details, while ignoring potentially useful information in the scene. In this work, we introduce a simple, yet novel, method: "Image Captioning by Committee Consensus" ($IC^3$), designed to generate a single caption that captures high-level details from several viewpoints. Notably, humans rate captions produced by $IC^3$ at least as helpful as baseline SOTA models more than two thirds of the time, and $IC^3$ captions can improve the performance of SOTA automated recall systems by up to 84%, indicating significant material improvements over existing SOTA approaches for visual description. Our code is publicly available at https://github.com/DavidMChan/caption-by-committee
Fully decentralized learning, where the global information, i.e., the actions of other agents, is inaccessible, is a fundamental challenge in cooperative multi-agent reinforcement learning. However, the convergence and optimality of most decentralized algorithms are not theoretically guaranteed, since the transition probabilities are non-stationary as all agents are updating policies simultaneously. To tackle this challenge, we propose best possible operator, a novel decentralized operator, and prove that the policies of agents will converge to the optimal joint policy if each agent independently updates its individual state-action value by the operator. Further, to make the update more efficient and practical, we simplify the operator and prove that the convergence and optimality still hold with the simplified one. By instantiating the simplified operator, the derived fully decentralized algorithm, best possible Q-learning (BQL), does not suffer from non-stationarity. Empirically, we show that BQL achieves remarkable improvement over baselines in a variety of cooperative multi-agent tasks.
In this work, we study the problem of real-time tracking and reconstruction of an information source with the purpose of actuation. A device monitors an $N$-state Markov process and transmits status updates to a receiver over a wireless erasure channel. We consider a set of joint sampling and transmission policies, including a semantics-aware one, and we study their performance with respect to relevant metrics. Specifically, we investigate the real-time reconstruction error and its variance, the consecutive error, the cost of memory error, and the cost of actuation error. Furthermore, we propose a randomized stationary sampling and transmission policy and derive closed-form expressions for all aforementioned metrics. We then formulate an optimization problem for minimizing the real-time reconstruction error subject to a sampling cost constraint. Our results show that in the scenario of constrained sampling generation, the optimal randomized stationary policy outperforms all other sampling policies when the source is rapidly evolving. Otherwise, the semantics-aware policy performs the best.
Recent studies have proven that DNNs, unlike human vision, tend to exploit texture information rather than shape. Such texture bias is one of the factors for the poor generalization performance of DNNs. We observe that the texture bias negatively affects not only in-domain generalization but also out-of-distribution generalization, i.e., Domain Generalization. Motivated by the observation, we propose a new framework to reduce the texture bias of a model by a novel optimization-based data augmentation, dubbed Stylized Dream. Our framework utilizes adaptive instance normalization (AdaIN) to augment the style of an original image yet preserve the content. We then adopt a regularization loss to predict consistent outputs between Stylized Dream and original images, which encourages the model to learn shape-based representations. Extensive experiments show that the proposed method achieves state-of-the-art performance in out-of-distribution settings on public benchmark datasets: PACS, VLCS, OfficeHome, TerraIncognita, and DomainNet.
Over the past few decades, with the rapid development of global aerospace and aerial remote sensing technology, the types of sensors have evolved from the traditional monomodal sensors (e.g., optical sensors) to the new generation of multimodal sensors [e.g., multispectral, hyperspectral, light detection and ranging (LiDAR) and synthetic aperture radar (SAR) sensors]. These advanced devices can dynamically provide various and abundant multimodal remote sensing images with different spatial, temporal, and spectral resolutions according to different application requirements. Since then, it is of great scientific significance to carry out the research of multimodal remote sensing image registration, which is a crucial step for integrating the complementary information among multimodal data and making comprehensive observations and analysis of the Earths surface. In this work, we will present our own contributions to the field of multimodal image registration, summarize the advantages and limitations of existing multimodal image registration methods, and then discuss the remaining challenges and make a forward-looking prospect for the future development of the field.
Understanding the extent to which the perceptual world can be recovered from language is a fundamental problem in cognitive science. We reformulate this problem as that of distilling psychophysical information from text and show how this can be done by combining large language models (LLMs) with a classic psychophysical method based on similarity judgments. Specifically, we use the prompt auto-completion functionality of GPT3, a state-of-the-art LLM, to elicit similarity scores between stimuli and then apply multidimensional scaling to uncover their underlying psychological space. We test our approach on six perceptual domains and show that the elicited judgments strongly correlate with human data and successfully recover well-known psychophysical structures such as the color wheel and pitch spiral. We also explore meaningful divergences between LLM and human representations. Our work showcases how combining state-of-the-art machine models with well-known cognitive paradigms can shed new light on fundamental questions in perception and language research.
We present an end-to-end binaural impulse response generator (BIR) to generate plausible sounds in real-time for real-world models. Our approach uses a novel neural-network-based BIR generator (Scene2BIR) for the reconstructed 3D model. We propose a graph neural network that uses both the material and the topology information of the 3D scenes and generates a scene latent vector. Moreover, we use a conditional generative adversarial network (CGAN) to generate BIRs from the scene latent vector. Our network is able to handle holes or other artifacts in the reconstructed 3D mesh model. We present an efficient cost function to the generator network to incorporate spatial audio effects. Given the source and the listener position, our approach can generate a BIR in 0.1 milliseconds on an NVIDIA GeForce RTX 2080 Ti GPU and can easily handle multiple sources. We have evaluated the accuracy of our approach with real-world captured BIRs and an interactive geometric sound propagation algorithm.
Should we care whether AI systems have representations of the world that are similar to those of humans? We provide an information-theoretic analysis that suggests that there should be a U-shaped relationship between the degree of representational alignment with humans and performance on few-shot learning tasks. We confirm this prediction empirically, finding such a relationship in an analysis of the performance of 491 computer vision models. We also show that highly-aligned models are more robust to both adversarial attacks and domain shifts. Our results suggest that human-alignment is often a sufficient, but not necessary, condition for models to make effective use of limited data, be robust, and generalize well.
Recently, test-time adaptation (TTA) has been proposed as a promising solution for addressing distribution shifts. It allows a base model to adapt to an unforeseen distribution during inference by leveraging the information from the batch of (unlabeled) test data. However, we uncover a novel security vulnerability of TTA based on the insight that predictions on benign samples can be impacted by malicious samples in the same batch. To exploit this vulnerability, we propose Distribution Invading Attack (DIA), which injects a small fraction of malicious data into the test batch. DIA causes models using TTA to misclassify benign and unperturbed test data, providing an entirely new capability for adversaries that is infeasible in canonical machine learning pipelines. Through comprehensive evaluations, we demonstrate the high effectiveness of our attack on multiple benchmarks across six TTA methods. In response, we investigate two countermeasures to robustify the existing insecure TTA implementations, following the principle of "security by design". Together, we hope our findings can make the community aware of the utility-security tradeoffs in deploying TTA and provide valuable insights for developing robust TTA approaches.
Systems with 1-bit quantization and oversampling are promising for the Internet of Things (IoT) devices in order to reduce the power consumption of the analog-to-digital-converters. The novel time-instance zero-crossing (TI ZX) modulation is a promising approach for this kind of channels but existing studies rely on optimization problems with high computational complexity and delay. In this work, we propose a practical waveform design based on the established TI ZX modulation for a multiuser multi-input multi-output (MIMO) downlink scenario with 1-bit quantization and temporal oversampling at the receivers. In this sense, the proposed temporal transmit signals are constructed by concatenating segments of coefficients which convey the information into the time-instances of zero-crossings according to the TI ZX mapping rules. The proposed waveform design is compared with other methods from the literature. The methods are compared in terms of bit error rate and normalized power spectral density. Numerical results show that the proposed technique is suitable for multiuser MIMO system with 1-bit quantization while tolerating some small amount of out-of-band radiation.