The high efficiency in computation and storage makes hashing (including binary hashing and quantization) a common strategy in large-scale retrieval systems. To alleviate the reliance on expensive annotations, unsupervised deep hashing becomes an important research problem. This paper provides a novel solution to unsupervised deep quantization, namely Contrastive Quantization with Code Memory (MeCoQ). Different from existing reconstruction-based strategies, we learn unsupervised binary descriptors by contrastive learning, which can better capture discriminative visual semantics. Besides, we uncover that codeword diversity regularization is critical to prevent contrastive learning-based quantization from model degeneration. Moreover, we introduce a novel quantization code memory module that boosts contrastive learning with lower feature drift than conventional feature memories. Extensive experiments on benchmark datasets show that MeCoQ outperforms state-of-the-art methods.

Dearth of labeled data for training supervised deep learning models for real-world applications of computer vision plagues the large-scale deployment of machine learning models in many domains; including problems in geospatial analysis and remote sensing. As an example, detecting infrequent events or changes like road closures, road blocks, junctions changing to roundabouts, temporary turn restrictions, etc. are critical to keep a geospatial mapping service up-to-date in real-time and can significantly improve user experience and above all, user safety. Obtaining labels to train supervised models to detect infrequent mobility change events is expensive in time and money for a multitude of reasons [13]. An inexpensive solution is synthetic generation of training data and labels based on user-provided conditional inputs that can be easily manipulated. In [13], we propose a novel deep conditional generative model that can synthetically generate various types of semantically rich, imagelike representations of GPS trajectory data (e.g., CRM, HCRM discussed below) by conditioning simultaneously on pixel-level (e.g, road network) and feature-level (e.g., desired observation time interval) conditional inputs. Detection (e.g., pedestrian crosswalks, road centerlines) and classification (e.g, landcover, vegetation) of geospatial features are routinely cast as canonical tasks in computer vision such as object detection, semantic segmentation, instance segmentation, etc. Analogously, detecting locations affected by the aforementioned changes in mobility can be cast as a semantic segmentation task from image-like representations of privacy-preserving GPS trajectory datasets [13]. Consider a raster representation of the earth's surface created with zoom-24 tiles [3]. Any contiguous set of n n tiles can be considered as an n n image whose pixels are the associated zoom-24 tiles. A count-based raster map (CRM) is a single-channel image-like representation where the value of each pixel is the number of GPS trace occurrences in the zoom-24 tile corresponding to the pixel, counted over all trajectories during an observation time interval, τ t.

A multi-turn dialogue always follows a specific topic thread, and topic shift at the discourse level occurs naturally as the conversation progresses, necessitating the model's ability to capture different topics and generate topic-aware responses. Previous research has either predicted the topic first and then generated the relevant response, or simply applied the attention mechanism to all topics, ignoring the joint distribution of the topic prediction and response generation models and resulting in uncontrollable and unrelated responses. In this paper, we propose a joint framework with a topic refinement mechanism to learn these two tasks simultaneously. Specifically, we design a three-pass iteration mechanism to generate coarse response first, then predict corresponding topics, and finally generate refined response conditioned on predicted topics. Moreover, we utilize GPT2DoubleHeads and BERT for the topic prediction task respectively, aiming to investigate the effects of joint learning and the understanding ability of GPT model. Experimental results demonstrate that our proposed framework achieves new state-of-the-art performance at response generation task and the great potential understanding capability of GPT model.

This paper demonstrates that by fine-tuning an autoregressive language model (GPT-Neo[1], [2]) on appropriately structured step-by-step demonstrations, it is possible to teach it to execute a mathematical task that has previously proved difficult for Transformers - longhand modulo operations - with a relatively small number of examples. Specifically, we fine-tune GPT-Neo to solve the numbers__div_remainder task from the DeepMind Mathematics Dataset; Saxton et al. [3] reported below 40% accuracy on this task with 2 million training examples. We show that after fine-tuning on 200 appropriately structured demonstrations of solving long division problems and reporting the remainders, the smallest available GPT-Neo model achieves over 80% accuracy. This is achieved by constructing an appropriate dataset for fine-tuning, with no changes to the learning algorithm. These results suggest that fine-tuning autoregressive language models on small sets of well-crafted demonstrations may be a useful paradigm for enabling individuals without training in machine learning to coax such models to perform some kinds of complex multi-step tasks.

Adversarial training, a method for learning robust deep neural networks, constructs adversarial examples during training. However, recent methods for generating NLP adversarial examples involve combinatorial search and expensive sentence encoders for constraining the generated instances. As a result, it remains challenging to use vanilla adversarial training to improve NLP models' performance, and the benefits are mainly uninvestigated. This paper proposes a simple and improved vanilla adversarial training process for NLP models, which we name Attacking to Training (A2T). The core part of A2T is a new and cheaper word substitution attack optimized for vanilla adversarial training. We use A2T to train BERT and RoBERTa models on IMDB, Rotten Tomatoes, Yelp, and SNLI datasets. Our results empirically show that it is possible to train robust NLP models using a much cheaper adversary. We demonstrate that vanilla adversarial training with A2T can improve an NLP model's robustness to the attack it was originally trained with and also defend the model against other types of word substitution attacks. Furthermore, we show that A2T can improve NLP models' standard accuracy, cross-domain generalization, and interpretability. Code is available at https://github.com/QData/Textattack-A2T .

Causal inference using observational text data is becoming increasingly popular in many research areas. This paper presents the Bayesian Topic Regression (BTR) model that uses both text and numerical information to model an outcome variable. It allows estimation of both discrete and continuous treatment effects. Furthermore, it allows for the inclusion of additional numerical confounding factors next to text data. To this end, we combine a supervised Bayesian topic model with a Bayesian regression framework and perform supervised representation learning for the text features jointly with the regression parameter training, respecting the Frisch-Waugh-Lovell theorem. Our paper makes two main contributions. First, we provide a regression framework that allows causal inference in settings when both text and numerical confounders are of relevance. We show with synthetic and semi-synthetic datasets that our joint approach recovers ground truth with lower bias than any benchmark model, when text and numerical features are correlated. Second, experiments on two real-world datasets demonstrate that a joint and supervised learning strategy also yields superior prediction results compared to strategies that estimate regression weights for text and non-text features separately, being even competitive with more complex deep neural networks.

Traffic flow forecasting is a crucial task in urban computing. The challenge arises as traffic flows often exhibit intrinsic and latent spatio-temporal correlations that cannot be identified by extracting the spatial and temporal patterns of traffic data separately. We argue that such correlations are universal and play a pivotal role in traffic flow. We put forward spacetime interval learning as a paradigm to explicitly capture these correlations through a unified analysis of both spatial and temporal features. Unlike the state-of-the-art methods, which are restricted to a particular road network, we model the universal spatio-temporal correlations that are transferable from cities to cities. To this end, we propose a new spacetime interval learning framework that constructs a local-spacetime context of a traffic sensor comprising the data from its neighbors within close time points. Based on this idea, we introduce spacetime neural network (STNN), which employs novel spacetime convolution and attention mechanism to learn the universal spatio-temporal correlations. The proposed STNN captures local traffic patterns, which does not depend on a specific network structure. As a result, a trained STNN model can be applied on any unseen traffic networks. We evaluate the proposed STNN on two public real-world traffic datasets and a simulated dataset on dynamic networks. The experiment results show that STNN not only improves prediction accuracy by 15% over state-of-the-art methods, but is also effective in handling the case when the traffic network undergoes dynamic changes as well as the superior generalization capability.

Compared with commonly used clinical risk factors, a sophisticated type of artificial intelligence (AI) called deep learning does a better job distinguishing between the mammograms of women who will later develop breast cancer and those who will not, according to a new study in the journal Radiology. Researchers said the findings underscore AI's potential as a second reader for radiologists that can reduce unnecessary imaging and associated costs. Annual mammography is recommended for women starting at age 40 to screen for breast cancer. Research has shown that screening mammography lowers breast cancer mortality by reducing the incidence of advanced cancer. Mammograms not only help detect cancer but also provide a measure of breast cancer risk through measurements of breast density.

Zodibot has expanded its A.I. expertise from horoscopes to visual art, and is even illustrating its own writing. The text in the following piece was generated by GPT-3 (trained on nanofiction by its human collaborator), and then programmatically illustrated by CLIP VQGAN.

In this article, we've listed all Machine Learning and Deep Learning courses by Andrew Ng, an excellent teacher from Standford University, and a tech-entrepreneur. The Machine Learning and Deep Learning courses given below are all available on Coursera in case you are interested in enrolling in any one of them. The Machine Learning course by Stanford and popularized by Andrew Ng's teaching is the best certification course in Machine Learning you can go for. The course is 11 weeks long and covers almost everything that you need to know about Machine Learning with great examples and assignments. The course has a 4.9/5 average rating from over 160,000 student ratings.