State-of-the-art vision and vision-and-language models rely on large-scale visio-linguistic pretraining for obtaining good performance on a variety of downstream tasks. Generally, such models are often either cross-modal (contrastive) or multi-modal (with earlier fusion) but not both; and they often only target specific modalities or tasks. A promising direction would be to use a single holistic universal model, as a “foundation”, that targets all modalities at once-a true vision and language foundation model should be good at vision tasks, language tasks, and cross- and multi-modal vision and language tasks. We introduce FLAVA as such a model and demonstrate impressive performance on a wide range of 35 tasks spanning these target modalities.

This paper studies a simple extension of image-based Masked Autoencoders (MAE) to self-supervised representation learning from audio spectrograms. Following the Transformer encoder-decoder design in MAE, our Audio-MAE first encodes audio spectrogram patches with a high masking ratio, feeding only the non-masked tokens through encoder layers. The decoder then re-orders and decodes the encoded context padded with mask tokens, in order to reconstruct the input spectrogram. We find it beneficial to incorporate local window attention in the decoder, as audio spectrograms are highly correlated in local time and frequency bands. We then fine-tune the encoder with a lower masking ratio on target datasets. Empirically, Audio-MAE sets new state-of-the-art performance on six audio and speech classification tasks, outperforming other recent models that use external supervised pre-training. Our code and models is available at https://github.com/facebookresearch/AudioMAE.

A crucial component for the scene text based reasoning required for TextVQA and TextCaps datasets involve detecting and recognizing text present in the images using an optical character recognition (OCR) system. The current systems are crippled by the unavailability of ground truth text annotations for these datasets as well as lack of scene text detection and recognition datasets on real images disallowing the progress in the field of OCR and evaluation of scene text based reasoning in isolation from OCR systems. In this work, we propose TextOCR, an arbitrary-shaped scene text detection and recognition with 900k annotated words collected on real images from TextVQA dataset. We show that current state-of-the-art text-recognition (OCR) models fail to perform well on TextOCR and that training on TextOCR helps achieve state-of-the-art performance on multiple other OCR datasets as well. We use a TextOCR trained OCR model to create PixelM4C model which can do scene text based reasoning on an image in an end-to-end fashion, allowing us to revisit several design choices to achieve new state-of-the-art performance on TextVQA dataset.

Can general-purpose neural models learn to navigate? For PointGoal navigation (‘go to Δx, Δy’), the answer is a clear ‘yes’ – mapless neural models composed of task-agnostic components (CNNs and RNNs) trained with large-scale model-free reinforcement learning achieve near-perfect performance [27]. However, for ObjectGoal navigation (‘find a TV’), this is an open question; one we tackle in this paper. The current best-known result on ObjectNav with general-purpose models is 6% success rate [25].First, we show that the key problem is overfitting. Large-scale training results in 94% success rate on training environments and only 8% in validation. We observe that this stems from agents memorizing environment layouts during training – sidestepping the need for exploration and directly learning shortest paths to nearby goal objects. We show that this is a natural consequence of optimizing for the task metric (which in fact penalizes exploration), is enabled by powerful observation encoders, and is possible due to the finite set of training environment configurations.Informed by our findings, we introduce Treasure Hunt Data Augmentation (THDA) to address overfitting in ObjectNav. THDA inserts 3D scans of household objects at arbitrary scene locations and uses them as ObjectNav goals – augmenting and greatly expanding the set of training layouts. Taken together with our other proposed changes, we improve the state of art on the Habitat ObjectGoal Navigation benchmark by 90% (from 14% success rate to 27%) and path efficiency by 48% (from 7.5 SPL to 11.1 SPL).

Performance on the most commonly used Visual Question Answering dataset (VQA v2) is starting to approach human accuracy. However, in interacting with state-of-the-art VQA models, it is clear that the problem is far from being solved. In order to stress test VQA models, we benchmark them against human-adversarial examples. Human subjects interact with a state-of-the-art VQA model, and for each image in the dataset, attempt to find a question where the model’s predicted answer is incorrect. We find that a wide range of state-of-the-art models perform poorly when evaluated on these examples. We conduct an extensive analysis of the collected adversarial examples and provide guidance on future research directions. We hope that this Adversarial VQA (AdVQA) benchmark can help drive progress in the field and advance the state of the art.

We introduce Habitat 2.0 (H2.0), a simulation platform for training virtual robots in interactive 3D environments and complex physics-enabled scenarios. We make comprehensive contributions to all levels of the embodied AI stack – data, simulation, and benchmark tasks. Specifically, we present: (i) ReplicaCAD: an artist-authored, annotated, reconfigurable 3D dataset of apartments (matching real spaces) with articulated objects (e.g. cabinets and drawers that can open/close); (ii) H2.0: a high-performance physics-enabled 3D simulator with speeds exceeding 25,000 simulation steps per second (850x real-time) on an 8-GPU node, representing 100x speed-ups over prior work; and, (iii) Home Assistant Benchmark (HAB): a suite of common tasks for assistive robots (tidy the house, stock groceries, set the table) that test a range of mobile manipulation capabilities. These large-scale engineering contributions allow us to systematically compare deep reinforcement learning (RL) at scale and classical sense-plan-act (SPA) pipelines in long-horizon structured tasks, with an emphasis on generalization to new objects, receptacles, and layouts. We find that (1) flat RL policies struggle on HAB compared to hierarchical ones; (2) a hierarchy with independent skills suffers from ‘hand-off problems’, and (3) SPA pipelines are more brittle than RL policies.

The ever-increasing amount of information flowing through Social Media forces the members of these networks to compete for attention and influence by relying on other people to spread their message. A large study of information propagation within Twitter reveals that the majority of users act as passive information consumers and do not forward the content to the network. Therefore, in order for individuals to become influential they must not only obtain attention and thus be popular, but also overcome user passivity. We propose an algorithm that determines the influence and passivity of users based on their information forwarding activity. An evaluation performed with a 2.5 million user dataset shows that our influence measure is a good predictor of URL clicks, outperforming several other measures that do not explicitly take user passivity into account. We demonstrate that high popularity does not necessarily imply high influence and vice-versa. These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

In peer-to-peer content delivery systems, such as BitTorrent, there may exist nodes that are non-cooperative and do not contribute their upload bandwidth to the system while still downloading content from others. The current widely used countermeasures against this freeriding behavior have been shown to be ineffective. In this paper, we address the problem by leveraging the trust latent in the social networks and explicitly incorporating the social links as part of the BitTorrent content distribution infrastructure. Our extensive system evaluation produces several insights. First, the social network topology alone without the trackers is an efficient and scalable content distribution medium. Second, thanks to the cooperative social links, BitTorrent's robustness to freeriding significantly improves. Finally, we find that a hybrid solution in which peers download from both their friends and other peers obtained from the trackers has the highest robustness to freeriding, shortest download completion times and the most balanced upload bandwidth utilization.

Wireless ad hoc networks are inherently vulnerable, as any node can disrupt the communication of potentially any other node in the network. Many solutions to this problem have been proposed. In this paper, we take a fresh and comprehensive approach that addresses simultaneously three aspects: security, scalability and adaptability to changing network conditions. Our communication protocol, Castor, occupies a unique point in the design space: it does not use any control messages except simple packet acknowledgments, and each node makes routing decisions locally and independently without exchanging any routing state with other nodes. Its novel design makes Castor resilient to a wide range of attacks and allows the protocol to scale to large network sizes and to remain efficient under high mobility. We compare Castor against four representative protocols from the literature. Our protocol achieves up to two times higher packet delivery rates, particularly in large and highly volatile networks, while incurring no or only limited additional overhead. At the same time, Castor is able to survive more severe attacks and recovers from them faster.

In ProtoPeer, the developer can switch between the simulation of a P2P system to its deployment on the actual network without changing a single line of code. This dramatically speeds up the implement-evaluate-reimplement cycle and allows for rapid system prototyping. Most of the major bugs and performance problems are caught early on during the simulation while the more time-consuming live deployment is used for the accurate evaluation of the final system.

Simulators are a commonly used tool in peer-to-peer systems research. However, they may not be able to capture all the details of a system operating in a live network deployment. Transitioning from simulation to the actual system implementation is a non-trivial and time-consuming task, a problem that we propose to solve in this paper. We present ProtoPeer, a peer-to-peer systems prototyping toolkit that allows for switching between the event-driven simulation and live network deployment without changing any of the application code. ProtoPeer exports a set of APIs for message passing, message queuing, timer operations as well as overlay routing and managing the overlay neighbors. Users can plug in their own custom implementations of most of the parts of ProtoPeer including custom network models for simulation and custom message passing over transports other than the default TCP/UDP.