Evolution Strategy (ES) algorithms have shown promising results in training complex robotic control policies due to their massive parallelism capability, simple implementation, effective parameter-space exploration, and fast training time. However, a key limitation of ES is its scalability to large capacity models, including modern neural network architectures. In this work, we develop Predictive Information Augmented Random Search (PI-ARS) to mitigate this limitation by leveraging recent advancements in representation learning to reduce the parameter search space for ES. Namely, PI-ARS combines a gradient-based representation learning technique, Predictive Information (PI), with a gradient-free ES algorithm, Augmented Random Search (ARS), to train policies that can process complex robot sensory inputs and handle highly nonlinear robot dynamics. We evaluate PI-ARS on a set of challenging visual-locomotion tasks where a quadruped robot needs to walk on uneven stepping stones, quincuncial piles, and moving platforms, as well as to complete an indoor navigation task. Across all tasks, PI-ARS demonstrates significantly better learning efficiency and performance compared to the ARS baseline. We further validate our algorithm by demonstrating that the learned policies can successfully transfer to a real quadruped robot, for example, achieving a 100% success rate on the real-world stepping stone environment, dramatically improving prior results achieving 40% success.

Chip floorplanning is a complex task within the physical design process, with more than six decades of research dedicated to it. In a recent paper published in Nature~\citemirhoseini2021graph, a new methodology based on deep reinforcement learning was proposed that solves the floorplanning problem for advanced chip technologies with production quality results. The proposed method enables generalization, which means that the quality of placements improves as the policy is trained on a larger number of chip blocks. In this paper, we describe Circuit Training, an open-source distributed reinforcement learning framework that re-implements the proposed methodology in TensorFlow v2.x. We will explain the framework and discuss ways it can be extended to solve other important problems within physical design and more generally chip design. We also show new experimental results that demonstrate the scaling and generalization performance of Circuit Training.

A longstanding goal of the field of AI is a method for learning a highly capable, generalist agent from diverse experience. In the subfields of vision and language, this was largely achieved by scaling up transformer-based models and training them on large, diverse datasets. Motivated by this progress, we investigate whether the same strategy can be used to produce generalist reinforcement learning agents. Specifically, we show that a single transformer-based model – with a single set of weights – trained purely offline can play a suite of up to 46 Atari games simultaneously at close-to-human performance. When trained and evaluated appropriately, we find that the same trends observed in language and vision hold, including scaling of performance with model size and rapid adaptation to new games via fine-tuning. We compare several approaches in this multi-game setting, such as online and offline RL methods and behavioral cloning, and find that our Multi-Game Decision Transformer models offer the best scalability and performance. We release the pre-trained models and code to encourage further research in this direction.

Learning effective visual representations that generalize well without human supervision is a fundamental problem in order to apply Machine Learning to a wide variety of tasks. Recently, two families of self-supervised methods, contrastive learning and latent bootstrapping, exemplified by SimCLR and BYOL respectively, have made significant progress. In this work, we hypothesize that adding explicit information compression to these algorithms yields better and more robust representations. We verify this by developing SimCLR and BYOL formulations compatible with the Conditional Entropy Bottleneck (CEB) objective, allowing us to both measure and control the amount of compression in the learned representation, and observe their impact on downstream tasks. Furthermore, we explore the relationship between Lipschitz continuity and compression, showing a tractable lower bound on the Lipschitz constant of the encoders we learn. As Lipschitz continuity is closely related to robustness, this provides a new explanation for why compressed models are more robust. Our experiments confirm that adding compression to SimCLR and BYOL significantly improves linear evaluation accuracies and model robustness across a wide range of domain shifts. In particular, the compressed version of BYOL achieves 76.0% Top-1 linear evaluation accuracy on ImageNet with ResNet-50, and 78.8% with ResNet-50 2x.

The Predictive Information is the mutual information between the past and the future, I(Xpast; Xfuture). We hypothesize that capturing the predictive information is useful in RL, since the ability to model what will happen next is necessary for success on many tasks. To test our hypothesis, we train Soft Actor-Critic (SAC) agents from pixels with an auxiliary task that learns a compressed representation of the predictive information of the RL environment dynamics using a contrastive version of the Conditional Entropy Bottleneck (CEB) objective. We refer to these as Predictive Information SAC (PI-SAC) agents. We show that PI-SAC agents can substantially improve sample efficiency over challenging baselines on tasks from the DM Control suite of continuous control environments. We evaluate PI-SAC agents by comparing against uncompressed PI-SAC agents, other compressed and uncompressed agents, and SAC agents directly trained from pixels. Our implementation is given on GitHub.

The goal of this paper is to serve as a guide for selecting a detection architecture that achieves the right speed/memory/accuracy balance for a given application and platform. To this end, we investigate various ways to trade accuracy for speed and memory usage in modern convolutional object detection systems. A number of successful systems have been proposed in recent years, but apples-toapples comparisons are difficult due to different base feature extractors (e.g., VGG, Residual Networks), different default image resolutions, as well as different hardware and software platforms. We present a unified implementation of the Faster R-CNN [30], R-FCN [6] and SSD [25] systems, which we view as meta-architectures and trace out the speed/accuracy trade-off curve created by using alternative feature extractors and varying other critical parameters such as image size within each of these meta-architectures. On one extreme end of this spectrum where speed and memory are critical, we present a detector that achieves real time speeds and can be deployed on a mobile device. On the opposite end in which accuracy is critical, we present a detector that achieves state-of-the-art performance measured on the COCO detection task.

Current image captioning methods are usually trained via maximum likelihood estimation. However, the log-likelihood score of a caption does not correlate well with human assessments of quality. Standard syntactic evaluation metrics, such as BLEU, METEOR and ROUGE, are also not well correlated. The newer SPICE and CIDEr metrics are better correlated, but have traditionally been hard to optimize for. In this paper, we show how to use a policy gradient (PG) method to directly optimize a linear combination of SPICE and CIDEr (a combination we call SPIDEr): the SPICE score ensures our captions are semantically faithful to the image, while CIDEr score ensures our captions are syntactically fluent. The PG method we propose improves on the prior MIXER approach, by using Monte Carlo rollouts instead of mixing MLE training with PG. We show empirically that our algorithm leads to easier optimization and improved results compared to MIXER. Finally, we show that using our PG method we can optimize any of the metrics, including the proposed SPIDEr metric which results in image captions that are strongly preferred by human raters compared to captions generated by the same model but trained to optimize MLE or the COCO metrics.

Models based on deep convolutional networks have dominated recent image interpretation tasks; we investigate whether models which are also recurrent are effective for tasks involving sequences, visual and otherwise. We describe a class of recurrent convolutional architectures which is end-to-end trainable and suitable for large-scale visual understanding tasks, and demonstrate the value of these models for activity recognition, image captioning, and video description. In contrast to previous models which assume a fixed visual representation or perform simple temporal averaging for sequential processing, recurrent convolutional models are "doubly deep" in that they learn compositional representations in space and time. Learning long-term dependencies is possible when nonlinearities are incorporated into the network state updates. Differentiable recurrent models are appealing in that they can directly map variable-length inputs (e.g., videos) to variable-length outputs (e.g., natural language text) and can model complex temporal dynamics; yet they can be optimized with backpropagation. Our recurrent sequence models are directly connected to modern visual convolutional network models and can be jointly trained to learn temporal dynamics and convolutional perceptual representations. Our results show that such models have distinct advantages over state-of-the-art models for recognition or generation which are separately defined or optimized.

In this paper we propose a technique to adapt convolutional neural network (CNN) based object detectors trained on RGB images to effectively leverage depth images at test time to boost detection performance. Given labeled depth images for a handful of categories we adapt an RGB object detector for a new category such that it can now use depth images in addition to RGB images at test time to produce more accurate detections. Our approach is built upon the observation that lower layers of a CNN are largely task and category agnostic and domain specific while higher layers are largely task and category specific while being domain agnostic. We operationalize this observation by proposing a mid-level fusion of RGB and depth CNNs. Experimental evaluation on the challenging NYUD2 dataset shows that our proposed adaptation technique results in an average 21% relative improvement in detection performance over an RGB-only baseline even when no depth training data is available for the particular category evaluated. We believe our proposed technique will extend advances made in computer vision to RGB-D data leading to improvements in performance at little additional annotation effort.

A major barrier towards scaling visual recognition systems is the difficulty of obtaining labeled images for large numbers of categories. Recently, deep convolutional neural networks (CNNs) trained used 1.2M+ labeled images have emerged as clear winners on object classification benchmarks. Unfortunately, only a small fraction of those labels are available with bounding box localization for training the detection task and even fewer pixel level annotations are available for semantic segmentation. It is much cheaper and easier to collect large quantities of image-level labels from search engines than it is to collect scene-centric images with precisely localized labels. We develop methods for learning large scale recognition models which exploit joint training over both weak (image-level) and strong (bounding box) labels and which transfer learned perceptual representations from strongly-labeled auxiliary tasks. We provide a novel formulation of a joint multiple instance learning method that includes examples from object-centric data with image-level labels when available, and also performs domain transfer learning to improve the underlying detector representation. We then show how to use our large scale detectors to produce pixel level annotations. Using our method, we produce a $>$7.6K category detector and release code and models at lsda.berkeley vision.org.

Sparse Filtering is a popular feature learning algorithm for image classification pipelines. In this paper, we connect the performance of Sparse Filtering with spectral properties of the corresponding feature matrices. This connection provides new insights into Sparse Filtering; in particular, it suggests early stopping of Sparse Filtering. We therefore introduce the Optimal Roundness Criterion (ORC), a novel stopping criterion for Sparse Filtering. We show that this stopping criterion is related with pre-processing procedures such as Statistical Whitening and demonstrate that it can make image classification with Sparse Filtering considerably faster and more accurate.

Models based on deep convolutional networks have dominated recent image interpretation tasks; we investigate whether models which are also recurrent, or “temporally deep”, are effective for tasks involving sequences, visual and otherwise. We develop a novel recurrent convolutional architecture suitable for large-scale visual learning which is end-to-end trainable, and demonstrate the value of these models on benchmark video recognition tasks, image description and retrieval problems, and video narration challenges. In contrast to current models which assume a fixed spatio-temporal receptive field or simple temporal averaging for sequential processing, recurrent convolutional models are “doubly deep” in that they can be compositional in spatial and temporal “layers”. Such models may have advantages when target concepts are complex and/or training data are limited. Learning long-term dependencies is possible when nonlinearities are incorporated into the network state updates. Long-term RNN models are appealing in that they directly can map variable-length inputs (e.g., video frames) to variable length outputs (e.g., natural language text) and can model complex temporal dynamics; yet they can be optimized with backpropagation. Our recurrent long-term models are directly connected to modern visual convnet models and can be jointly trained to simultaneously learn temporal dynamics and convolutional perceptual representations. Our results show such models have distinct advantages over state-of-the-art models for recognition or generation which are separately defined and/or optimized.

We present a system which can recognize the contents of your meal from a single image, and then predict its nutritional contents, such as calories. The simplest version assumes that the user is eating at a restaurant for which we know the menu. In this case, we can collect images offline to train a multi-label classifier. At run time, we apply the classifier (running on your phone) to predict which foods are present in your meal, and we lookup the corresponding nutritional facts. We apply this method to a new dataset of images from 23 different restaurants, using a CNN-based classifier, significantly outperforming previous work. The more challenging setting works outside of restaurants. In this case, we need to estimate the size of the foods, as well as their labels. This requires solving segmentation and depth / volume estimation from a single image. We present CNN-based approaches to these problems, with promising preliminary results.

Caffe provides multimedia scientists and practitioners with a clean and modifiable framework for state-of-the-art deep learning algorithms and a collection of reference models. The framework is a BSD-licensed C++ library with Python and MATLAB bindings for training and deploying general-purpose convolutional neural networks and other deep models efficiently on commodity architectures. Caffe fits industry and internet-scale media needs by CUDA GPU computation, processing over 40 million images a day on a single K40 or Titan GPU (approx 2 ms per image). By separating model representation from actual implementation, Caffe allows experimentation and seamless switching among platforms for ease of development and deployment from prototyping machines to cloud environments.

We present a holistic data-driven technique that generates natural-language descriptions for videos. We combine the output of state-of-the-art object and activity detectors with "real-world' knowledge to select the most probable subject-verb-object triplet for describing a video. We show that this knowledge, automatically mined from web-scale text corpora, enhances the triplet selection algorithm by providing it contextual information and leads to a four-fold increase in activity identification. Unlike previous methods, our approach can annotate arbitrary videos without requiring the expensive collection and annotation of a similar training video corpus. We evaluate our technique against a baseline that does not use text-mined knowledge and show that humans prefer our descriptions 61% of the time.

Despite a recent push towards large-scale object recognition, activity recognition remains limited to narrow domains and small vocabularies of actions. In this paper, we tackle the challenge of recognizing and describing activities ``in-the-wild''. We present a solution that takes a short video clip and outputs a brief sentence that sums up the main activity in the video, such as the actor, the action and its object. Unlike previous work, our approach works on out-of-domain actions: it does not require training videos of the exact activity. If it cannot find an accurate prediction for a pre-trained model, it finds a less specific answer that is also plausible from a pragmatic standpoint. We use semantic hierarchies learned from the data to help to choose an appropriate level of generalization, and priors learned from Web-scale natural language corpora to penalize unlikely combinations of actors/actions/objects, we also use a Web-scale language model to ``fill in'' novel verbs, i.e. when the verb does not appear in the training set. We evaluate our method on a large YouTube corpus and demonstrate it is able to generate short sentence descriptions of video clips better than baseline approaches.

We propose a system for human-robot interaction that learns both models for spatial prepositions and for object recognition. Our system grounds the meaning of an input sentence in terms of visual percepts coming from the robot's sensors in order to send an appropriate command to the PR2 or respond to spatial queries. To perform this grounding, the system recognizes the objects in the scene, determines which spatial relations hold between those objects, and semantically parses the input sentence. The proposed system uses the visual and spatial information in conjunction with the semantic parse to interpret statements that refer to objects (nouns), their spatial relationships (prepositions), and to execute commands (actions). The semantic parse is inherently compositional, allowing the robot to understand complex commands that refer to multiple objects and relations such as: “Move the cup close to the robot to the area in front of the plate and behind the tea box”. Our system correctly parses 94% of the 210 online test sentences, correctly interprets 91% of the correctly parsed sentences, and correctly executes 89% of the correctly interpreted sentences.

The contribution of this paper is to provide a semantic model (using soft constraints) of the words used by web-users to describe objects in a language game; a game in which one user describes a selected object of those composing the scene (see figure 1), and another user has to guess which object has been described. The given description needs to be non ambiguous and accurate enough to allow other users to guess the described shape correctly. To build these semantic models the descriptions need to be analyzed to extract the syntax and words' classes used (see [1] for details). We have modeled the meaning of these descriptions using soft constraints as a way for grounding the meaning. The descriptions generated by the system took into account the context of the object to avoid ambiguous descriptions, and allowed users to guess the described object correctly 72% of the times.

People have a particular way to describe objects into a scene. Our work will be trying to learn these ways with their special characteristics and describe new images automatically. For that, we need to learn the syntax of these descriptions, and generate new syntax-correct phrases, the main reason for this paper. The final objective of this work will be describing new scenes automatically that could be understood by humans.

Since the concept of fuzzy set was defined in 1965 by Zadeh, numerous papers on fuzzy topics have been published and many of his seminal ideas have evolved in different directions. However, there is lack of natural and intuitive procedures to capture fuzzy perceptions from non-expert users. In this paper, we propose an innovative approach to solve this problem with a straightforward and repeatable mechanism based on a web gadget. The key benefit of this approach is that users obtain an immediate feedback thanks to the use of a visual spray pen metaphor. We also propose several algorithms to model these fuzzy perceptions with trapezoidal fuzzy set approximations. These algorithms are then evaluated though a real case study in order to test the usability and usefulness of the proposed gadget. One of the main conclusions extracted from this work suggests that the evaluation of the fitness of these approximations is intrinsically subjective, requiring further development and research.

The main idea of this paper consists on doing gesture recognition by means of prediction. Taking into account that a signal predictor will predict accurately future values of gestures of its class and inaccurately the values of the others, we can use the prediction error to classify the gestures. These predictors are implemented using neuro fuzzy systems. We call this approach prediction-error-classification approach (PEC) and this idea represents a different approach to solve the problem of gesture recognition in real time using inexpensive accelerometers. To validate this approach we have studied the impact of the number of training samples in the prediction error using cross-validation. We have also studied the impact of the number of training samples in the recognition rate, using again cross-validation. And to test the robustness and applicability in a real situation of this approach, we have repeated all the tests with a more realistic experiment.

To avoid self-contradictions in fuzzy set theory, it is needed to consider only fuzzy sets that are not self-contradictory and whose complements are also not self-contradictory. This paper deals with a subset F/sub nc/(E) in [0,1]/sup E/, where the above conditions are satisfied. A particular fuzzy set theory is defined in F/sub nc/(E) with adequate logical connectives. Finally, which classical properties hold in the new theory is studied.