The full power of human language-based communication cannot be realized without negation. All human languages have some form of negation. Despite this, negation remains a challenging phenomenon for current natural language understanding systems. To facilitate the future development of models that can process negation effectively, we present CONDAQA, the first English reading comprehension dataset which requires reasoning about the implications of negated statements in paragraphs. We collect paragraphs with diverse negation cues, then have crowdworkers ask questions about the implications of the negated statement in the passage. We also have workers make three kinds of edits to the passage -- paraphrasing the negated statement, changing the scope of the negation, and reversing the negation -- resulting in clusters of question-answer pairs that are difficult for models to answer with spurious shortcuts. CONDAQA features 14,182 question-answer pairs with over 200 unique negation cues and is challenging for current state-of-the-art models. The best performing model on CONDAQA (UnifiedQA-v2-3b) achieves only 42% on our consistency metric, well below human performance which is 81%. We release our dataset, along with fully-finetuned, few-shot, and zero-shot evaluations, to facilitate the development of future NLP methods that work on negated language.

We introduce the Block-Recurrent Transformer, which applies a transformer layer in a recurrent fashion along a sequence, and has linear complexity with respect to sequence length. Our recurrent cell operates on blocks of tokens rather than single tokens during training, and leverages parallel computation within a block in order to make efficient use of accelerator hardware. The cell itself is strikingly simple. It is merely a transformer layer: it uses self-attention and cross-attention to efficiently compute a recurrent function over a large set of state vectors and tokens. Our design was inspired in part by LSTM cells, and it uses LSTM-style gates, but it scales the typical LSTM cell up by several orders of magnitude. Our implementation of recurrence has the same cost in both computation time and parameter count as a conventional transformer layer, but offers dramatically improved perplexity in language modeling tasks over very long sequences. Our model out-performs a long-range Transformer XL baseline by a wide margin, while running twice as fast. We demonstrate its effectiveness on PG19 (books), arXiv papers, and GitHub source code. Our code has been released as open source.

Next Best View computation (NBV) is a long-standing problem in robotics, and consists in identifying the next most informative sensor position(s) for reconstructing a 3D object or scene efficiently and accurately. Like most current methods, we consider NBV prediction from a depth sensor like Lidar systems. Learning-based methods relying on a volumetric representation of the scene are suitable for path planning, but have lower accuracy than methods using a surface-based representation. However, the latter do not scale well with the size of the scene and constrain the camera to a small number of poses. To obtain the advantages of both representations, we show that we can maximize surface metrics by Monte Carlo integration over a volumetric representation. In particular, we propose an approach, SCONE, that relies on two neural modules: The first module predicts occupancy probability in the entire volume of the scene. Given any new camera pose, the second module samples points in the scene based on their occupancy probability and leverages a self-attention mechanism to predict the visibility of the samples. Finally, we integrate the visibility to evaluate the gain in surface coverage for the new camera pose. NBV is selected as the pose that maximizes the gain in total surface coverage. Our method scales to large scenes and handles free camera motion: It takes as input an arbitrarily large point cloud gathered by a depth sensor as well as camera poses to predict NBV. We demonstrate our approach on a novel dataset made of large and complex 3D scenes.

Recognizing who is speaking in a crowded scene is a key challenge towards the understanding of the social interactions going on within. Detecting speaking status from body movement alone opens the door for the analysis of social scenes in which personal audio is not obtainable. Video and wearable sensors make it possible recognize speaking in an unobtrusive, privacy-preserving way. When considering the video modality, in action recognition problems, a bounding box is traditionally used to localize and segment out the target subject, to then recognize the action taking place within it. However, cross-contamination, occlusion, and the articulated nature of the human body, make this approach challenging in a crowded scene. Here, we leverage articulated body poses for subject localization and in the subsequent speech detection stage. We show that the selection of local features around pose keypoints has a positive effect on generalization performance while also significantly reducing the number of local features considered, making for a more efficient method. Using two in-the-wild datasets with different viewpoints of subjects, we investigate the role of cross-contamination in this effect. We additionally make use of acceleration measured through wearable sensors for the same task, and present a multimodal approach combining both methods.

Counterfactual data augmentation (CDA) -- i.e., adding minimally perturbed inputs during training -- helps reduce model reliance on spurious correlations and improves generalization to out-of-distribution (OOD) data. Prior work on generating counterfactuals only considered restricted classes of perturbations, limiting their effectiveness. We present COunterfactual Generation via Retrieval and Editing (CORE), a retrieval-augmented generation framework for creating diverse counterfactual perturbations for CDA. For each training example, CORE first performs a dense retrieval over a task-related unlabeled text corpus using a learned bi-encoder and extracts relevant counterfactual excerpts. CORE then incorporates these into prompts to a large language model with few-shot learning capabilities, for counterfactual editing. Conditioning language model edits on naturally occurring data results in diverse perturbations. Experiments on natural language inference and sentiment analysis benchmarks show that CORE counterfactuals are more effective at improving generalization to OOD data compared to other DA approaches. We also show that the CORE retrieval framework can be used to encourage diversity in manually authored perturbations

Gait recognition is widely used in diversified practical applications. Currently, the most prevalent approach is to recognize human gait from RGB images, owing to the progress of computer vision technologies. Nevertheless, the perception capability of RGB cameras deteriorates in rough circumstances, and visual surveillance may cause privacy invasion. Due to the robustness and non-invasive feature of millimeter wave (mmWave) radar, radar-based gait recognition has attracted increasing attention in recent years. In this research, we propose a Hierarchical Dynamic Network (HDNet) for gait recognition using mmWave radar. In order to explore more dynamic information, we propose point flow as a novel point clouds descriptor. We also devise a dynamic frame sampling module to promote the efficiency of computation without deteriorating performance noticeably. To prove the superiority of our methods, we perform extensive experiments on two public mmWave radar-based gait recognition datasets, and the results demonstrate that our model is superior to existing state-of-the-art methods.

Music composition, like most art forms, has for a long time been a skill specific to human beings. Music composition has an intuitive side to it necessary to determine which pitches create harmorny together, what chords can be played after a certain note, or what note progressions are in violation of intrinsic musical theory. With the recent successes in neural network modeling of predictive natural behaviour and generative models, there have been good applications of modelling note progression probabilities for music generation. The two dominant approaches to neural music generation are adversarial training Sutskever et al. [2014], Liu and Randall [2016], Yang et al. [2017], Dong et al. [2018], and sequence-to-sequence recurrent networks Chung et al. [2014], Waite [2016], Weel [2017], each with its merits. Although Wave-form representations have been shown to be a viable way to generate audio not necessarily specific to music Oord et al. [2016a], it is symbolic representations that are favoured in literature for the task of music generation Mogren [2016], Yang et al. [2017], Lerdahl and Jackendoff [1983], Colombo and Gerstner [2018], Chung et al. [2014]. Owing to the existing lack of out-right comparisons between adversarial and non-adversarial training for music generation, the aim of this study is to compare music samples generated by two generative models, one trained in an adversarial setting, and the other in a non adversarial setting, using musical instrument digital interface (MIDI) data.

Retrieval-augmented language models (LMs) use non-parametric memory to substantially outperform their non-retrieval counterparts on perplexity-based evaluations, but it is an open question whether they achieve similar gains in few- and zero-shot end-task accuracy. We extensively study one such model, the k-nearest neighbor LM (kNN-LM), showing that the gains marginally transfer. The main challenge is to achieve coverage of the verbalizer tokens that define the different end-task class labels. To address this challenge, we also introduce kNN-Prompt, a simple and effective kNN-LM with automatically expanded fuzzy verbalizers (e.g. to expand terrible to also include silly and other task-specific synonyms for sentiment classification). Across nine diverse end-tasks, using kNN-Prompt with GPT-2 large yields significant performance boosts over strong zero-shot baselines (13.4% absolute improvement over the base LM on average). We also show that other advantages of non-parametric augmentation hold for end tasks; kNN-Prompt is effective for domain adaptation with no further training, and gains increase with the size of the retrieval model.

Deep learning (DL) finds rich applications in the wireless domain to improve spectrum awareness. Typically, DL models are either randomly initialized following a statistical distribution or pretrained on tasks from other domains in the form of transfer learning without accounting for the unique characteristics of wireless signals. Self-supervised learning (SSL) enables the learning of useful representations from Radio Frequency (RF) signals themselves even when only limited training data samples with labels are available. We present a self-supervised RF signal representation learning method and apply it to the automatic modulation recognition (AMR) task by specifically formulating a set of transformations to capture the wireless signal characteristics. We show that the sample efficiency (the number of labeled samples needed to achieve a certain performance) of AMR can be significantly increased (almost an order of magnitude) by learning signal representations with SSL. This translates to substantial time and cost savings. Furthermore, SSL increases the model accuracy compared to the state-of-the-art DL methods and maintains high accuracy when limited training data is available.

Zenfolio, the website builder and photo sharing site, recently introduced technology that applies AI to assist photographers (opens in new tab) in selecting the best photos from the thousands of shots typically taken during a photo session. The advanced image recognition technology is tremendously powerful and can make photographers more efficient than they ever dreamed. When exploring the best way to roll out AI to the photography community, Zenfolio had some fascinating discussions about corporate responsibility generally, and its obligations specifically. The concept of AI was theorized centuries ago by Greek philosophers (opens in new tab), with myths about Talos and Pandora creating chaos and destruction. Perhaps these cautionary tales about artificial beings influence our thinking today.