The data deluge comes with high demands for data labeling. Crowdsourcing (or, more generally, ensemble learning) techniques aim to produce accurate labels via integrating noisy, non-expert labeling from annotators. The classic Dawid-Skene estimator and its accompanying expectation maximization (EM) algorithm have been widely used, but the theoretical properties are not fully understood. Tensor methods were proposed to guarantee identification of the Dawid-Skene model, but the sample complexity is a hurdle for applying such approaches--since the tensor methods hinge on the availability of third-order statistics that are hard to reliably estimate given limited data. In this paper, we propose a framework using pairwise co-occurrences of the annotator responses, which naturally admits lower sample complexity. We show that the approach can identify the Dawid-Skene model under realistic conditions. We propose an algebraic algorithm reminiscent of convex geometry-based structured matrix factorization to solve the model identification problem efficiently, and an identifiability-enhanced algorithm for handling more challenging and critical scenarios. Experiments show that the proposed algorithms outperform the state-of-art algorithms under a variety of scenarios.

For completeness, in the following we include a datasheet based on the format of [1]. For what purpose was the dataset created? Was there a specific task in mind? Who created the dataset (e.g., which team, research group) and on behalf of which entity What do the instances that comprise the dataset represent (e.g., documents, photos, people, How many instances are there in total (of each type, if appropriate)? What data does each instance consist of?

Autonomous agents that control user interfaces to accomplish human tasks are emerging. Leveraging LLMs to power such agents has been of special interest, but unless fine-tuned on human-collected task demonstrations, performance is still relatively low. In this work we study whether fine-tuning alone is a viable approach for building real-world UI control agents.

Generic motion understanding from video involves not only tracking objects, but also perceiving how their surfaces deform and move. This information is useful to make inferences about 3D shape, physical properties and object interactions. While the problem of tracking arbitrary physical points on surfaces over longer video clips has received some attention, no dataset or benchmark for evaluation existed, until now.

To answer the first question, we split the data into the two groups: the first group contains the subset of data for numeric-simple prompts, and the second group the subset of data for attribute-color prompts. We only consider prompts in both groups that contain the same numbers (1-4) and the same words ("cat", "apple", "koala", "bottle", "mushroom"), to isolate the effect of adding the color term as opposed to potential confounding factors. For example a confounding factor might be the word identity, as a model might be more accurate in generating correct images when the prompt contains the word "dog", and if this word exists only in the first prompt type and not in the second then responses in the first prompt type will on average have higher accuracy that may or may not

This work proposes a new challenge set for multimodal classification, focusing on detecting hate speech in multimodal memes. It is constructed such that unimodal models struggle and only multimodal models can succeed: difficult examples ("benign confounders") are added to the dataset to make it hard to rely on unimodal signals. The task requires subtle reasoning, yet is straightforward to evaluate as a binary classification problem. We provide baseline performance numbers for unimodal models, as well as for multimodal models with various degrees of sophistication. We find that state-of-the-art methods perform poorly compared to humans, illustrating the difficulty of the task and highlighting the challenge that this important problem poses to the community.

Instruction following is one of the fundamental capabilities of large language models (LLMs). As the ability of LLMs is constantly improving, they have been increasingly applied to deal with complex human instructions in real-world scenarios. Therefore, how to evaluate the ability of complex instruction-following of LLMs has become a critical research problem. Existing benchmarks mainly focus on modeling different types of constraints in human instructions while neglecting the composition of different constraints, which is an indispensable constituent in complex instructions. To this end, we propose ComplexBench, a benchmark for comprehensively evaluating the ability of LLMs to follow complex instructions composed of multiple constraints. We propose a hierarchical taxonomy for complex instructions, including 4 constraint types, 19 constraint dimensions, and 4 composition types, and manually collect a high-quality dataset accordingly. To make the evaluation reliable, we augment LLM-based evaluators with rules to effectively verify whether generated texts can satisfy each constraint and composition. Furthermore, we obtain the final evaluation score based on the dependency structure determined by different composition types.

In this paper, we focus on training and evaluating effective word embeddings with both text and visual information. More specifically, we introduce a large-scale dataset with 300 million sentences describing over 40 million images crawled and downloaded from publicly available Pins (i.e. an image with sentence descriptions uploaded by users) on Pinterest [2]. This dataset is more than 200 times larger than MS COCO [22], the standard large-scale image dataset with sentence descriptions. In addition, we construct an evaluation dataset to directly assess the effectiveness of word embeddings in terms of finding semantically similar or related words and phrases. The word/phrase pairs in this evaluation dataset are collected from the click data with millions of users in an image search system, thus contain rich semantic relationships. Based on these datasets, we propose and compare several Recurrent Neural Networks (RNNs) based multimodal (text and image) models. Experiments show that our model benefits from incorporating the visual information into the word embeddings, and a weight sharing strategy is crucial for learning such multimodal embeddings. The project page is: http://www.stat.

We present a variety of new architectural features and training procedures that we apply to the generative adversarial networks (GANs) framework. Using our new techniques, we achieve state-of-the-art results in semi-supervised classification on MNIST, CIFAR-10 and SVHN. The generated images are of high quality as confirmed by a visual Turing test: our model generates MNIST samples that humans cannot distinguish from real data, and CIFAR-10 samples that yield a human error rate of 21.3%. We also present ImageNet samples with unprecedented resolution and show that our methods enable the model to learn recognizable features of ImageNet classes.

A.1 Annotation Process Prior to conducting the annotation process, human annotators undergo a comprehensive training phase. During this phase, annotators are encouraged to raise any doubts and present corner cases to the first author, who revisits the guidelines with further details and examples. Following the training phase, a meeting is held between the first author, the last author, and the annotation manager to ensure a clear understanding of the task. The manager then evaluates the performance of annotators before collecting the final annotations. Throughout the annotation process, the annotation manager, the first author, and the last author maintain regular communication, reviewing samples and addressing any potential issues such as the exclusion of videos with problematic or offensive content.