Search results

We propose simple active sampling and reweighting strategies for optimizing min-max fairness that can be applied to any classification or regression model learned via loss minimization. The key intuition behind our approach is to use at each timestep a datapoint from the group that is worst off under the current model for updating the model. The ease of implementation and the generality of our robust formulation

We introduce the “Shopping Queries Data Set”, a large dataset of difficult search queries, released with the aim of fostering research in the area of semantic matching of queries and products. For each query, the dataset provides a list of up to 40 potentially relevant results, together with ESCI relevance judgements (Exact, Substitute, Complement, Irrelevant) indicating the relevance of the product to

Here, we study the problem of causal generalization—generalizing from the observational to interventional distributions—in forecasting. Our goal is to find answers to the question: How does the efficacy of an autoregressive (VAR) model in predicting statistical associations compare with its ability to predict under interventions? To this end, we introduce the framework of causal learning theory for forecasting

Our datasets cover binary, multi-class, and multilabel classification tasks and represent various ways information is organized in a long text (e.g. information that is critical to making the classification decision is at the beginning or toward the end of the document). Our results show that more complex models often fail to outperform simple baselines and yield inconsistent performance across datasets

We study task-agnostic continual reinforcement learning (TACRL) in which standard RL challenges are compounded with partial observability stemming from ask agnosticism, as well as additional difficulties of continual learning (CL), i.e., learning on a non-stationary sequence of tasks. Here we compare TACRL methods with their soft upper bounds prescribed by previous literature: multi-task learning (MTL)

MASSIVE is a parallel dataset of > 1M utterances across 52 languages with annotations for the Natural Language Understanding tasks of intent prediction and slot annotation. Utterances span 60 intents and include 55 slot types. MASSIVE was created by localizing the Spoken Language Understanding Resource Package (SLURP) dataset, composed of general intelligent voice assistant single-shot interactions.

AlexaTM 20B is a 20B-Parameter sequence-to-sequence transformer model created by the Alexa Teacher Model (AlexaTM) team at Amazon. The model was trained on a mixture of Common Crawl (mC4) and Wikipedia data across 12 languages using denoising and Causal Language Modeling (CLM) tasks. AlexaTM 20B can be used for in-context learning. "In-context learning," also known as "prompting," refers to a method for

Inspired by two basic mechanisms in animal visual systems, we introduce a feature transform technique that imposes invariance properties in the training of deep neural networks. The resulting algorithm requires less parameter tuning, trains well with an initial learning rate 1.0, and easily generalizes to different tasks. We enforce scale invariance with local statistics in the data to align similar samples

Combinatorial optimization problems are pervasive across science and industry. Modern deep learning tools are poised to solve these problems at unprecedented scales, but a unifying framework that incorporates insights from statistical physics is still outstanding. Here we demonstrate how graph neural networks can be used to solve combinatorial optimization problems. Our approach is broadly applicable to

In recent years, the dominant paradigm for text spotting is to combine the tasks of text detection and recognition into a single end-to-end framework. Under this paradigm, both tasks are accomplished by operating over a shared global feature map extracted from the input image. Among the main challenges that end-to-end approaches face is the performance degradation when recognizing text across scale variations

We consider the problem of omni-supervised object detection, which can use unlabeled, fully labeled and weakly labeled annotations, such as image tags, counts, points, etc., for object detection. This is enabled by a unified architecture, Omni-DETR, based on the recent progress on student-teacher framework and end-to-end transformer based object detection. Under this unified architecture, different types

It's an end-to-end blueprint architecture for real-time fraud detection using graph database Amazon Neptune, Amazon SageMaker and Deep Graph Library (DGL) to construct a heterogeneous graph from tabular data and train a Graph Neural Network(GNN) model to detect fraudulent transactions in the IEEE-CIS Fraud detection dataset. See the more detail in this AWS blog post.

This repository contains data, relevant scripts and baseline code for the DSTC11 summer track on Intent Induction from Conversations for Task-Oriented Dialogue. This track aims to evaluate methods for the automatic induction of customer intents in the realistic setting of customer service interactions between human agents and customers. As complete conversations will be provided, participants can make use

We propose a memory efficient method, named Stochastic Backpropagation (SBP), for training deep neural networks on videos. It is based on the finding that gradients from incomplete execution for backpropagation can still effectively train the models with minimal accuracy loss, which attributes to the high redundancy of video. SBP keeps all forward paths but randomly and independently removes the backward

We present Syne Tune, a library for large-scale distributed hyperparameter optimization (HPO). Syne Tune’s modular architecture allows users to easily switch between different execution backends to facilitate experimentation and makes it easy to contribute new optimization algorithms. To foster reproducible benchmarking, Syne Tune provides an efficient simulator backend and a benchmarking suite, which are

This repository hosts the REDTab dataset, the largest natural-table relation extraction dataset. In our dataset, we have a total of 9,149 tables, and 22,236 column pairs from Music and Literature domains. We release two categories of REDTab: REDTab-standard: This is the standard REDTab which includes 23 different relations. REDTab-simple: This is a filtered version which only focuses on the top 10 most

Existing approaches for Structure from Motion (SfM) produce impressive 3-D reconstruction results especially when using imagery captured with large parallax. However, to create engaging video-content in movies and TV shows, the amount by which a camera can be moved while filming a particular shot is often limited. The resulting small-motion parallax between video frames makes standard geometry-based SfM

Research Service Workbench (RSW) on AWS is a backend API designed to empower AWS partners to more easily build self-service portals for research institutions, allowing their researchers to create and manage secure research computing environments. RSW builds on AWS services including Amazon S3, Service Catalog, and AWS Systems Manager to deliver comprehensive, cutting-edge capabilities in less time while

Class-incremental learning (CIL) has been widely studied under the setting of starting from a small number of classes (base classes). Instead, we explore an understudied real-world setting of CIL that starts with a strong model pre-trained on a large number of base classes. We hypothesize that a strong base model can provide a good representation for novel classes and incremental learning can be done with

Being able to spot defective parts is a critical component in large-scale industrial manufacturing. A particular challenge that we address in this work is the cold-start problem: fit a model using nominal (non-defective) example images only. While handcrafted solutions per class are possible, the goal is to build systems that work well simultaneously on many different tasks automatically. The best performing