联合学习（FL）是一种使用跨设备分布的数据训练模型的技术。差异隐私（DP）为敏感数据提供了正式的隐私保证。我们的目标是在使用FL和DP保护隐私的同时，在计算受限设备上训练大型神经网络语言模型（NNLM）。但是，随着模型大小的增长，引入模型的DP噪声增加，这通常会阻止收敛。我们提出了部分嵌入更新（PEU），这是一种新颖的技术，可以通过降低有效载荷大小来降低噪声。此外，我们采用低级适应（LORA）和噪声对比估计（NCE）来减少计算受限设备上大型模型的记忆需求。这种技术的组合使得可以在保留准确性和隐私的同时训练大型唱机语言模型。

联合学习是一种新兴的机器学习（ML）范式，其中大量设备集体训练ML模型，而数据仍保留在设备上。该研究领域有一系列独特的实践挑战，为了系统地取得进步，需要策划与此范式兼容的新数据集。图像域中的现有联合学习基准不能准确捕获许多实际用例的规模和异质性。我们介绍了Flair，这是一个具有挑战性的大规模注释图像数据集，用于适合联合学习的多标签分类。弗莱尔（Flair）拥有来自51,414个Flickr用户的429,078张图像，并捕获了联合学习中通常遇到的许多复杂性，例如异质用户数据和长尾标签分布。我们在此数据集上的不同任务中实现了不同的学习设置中的多个基线。我们认为，天赋可以作为推进联邦学习最先进的具有挑战性的基准。数据集访问和基准的代码可在\ url {https://github.com/apple/ml-flair}上获得。

Efficient and robust control using spiking neural networks (SNNs) is still an open problem. Whilst behaviour of biological agents is produced through sparse and irregular spiking patterns, which provide both robust and efficient control, the activity patterns in most artificial spiking neural networks used for control are dense and regular -- resulting in potentially less efficient codes. Additionally, for most existing control solutions network training or optimization is necessary, even for fully identified systems, complicating their implementation in on-chip low-power solutions. The neuroscience theory of Spike Coding Networks (SCNs) offers a fully analytical solution for implementing dynamical systems in recurrent spiking neural networks -- while maintaining irregular, sparse, and robust spiking activity -- but it's not clear how to directly apply it to control problems. Here, we extend SCN theory by incorporating closed-form optimal estimation and control. The resulting networks work as a spiking equivalent of a linear-quadratic-Gaussian controller. We demonstrate robust spiking control of simulated spring-mass-damper and cart-pole systems, in the face of several perturbations, including input- and system-noise, system disturbances, and neural silencing. As our approach does not need learning or optimization, it offers opportunities for deploying fast and efficient task-specific on-chip spiking controllers with biologically realistic activity.

Recent advances in language modeling have enabled new conversational systems. In particular, it is often desirable for people to make choices among specified options when using such systems. We address the problem of reference resolution, when people use natural expressions to choose between real world entities. For example, given the choice `Should we make a Simnel cake or a Pandan cake?' a natural response from a non-expert may be indirect: `let's make the green one'. Reference resolution has been little studied with natural expressions, thus robustly understanding such language has large potential for improving naturalness in dialog, recommendation, and search systems. We create AltEntities (Alternative Entities), a new public dataset of entity pairs and utterances, and develop models for the disambiguation problem. Consisting of 42K indirect referring expressions across three domains, it enables for the first time the study of how large language models can be adapted to this task. We find they achieve 82%-87% accuracy in realistic settings, which while reasonable also invites further advances.

The number of international benchmarking competitions is steadily increasing in various fields of machine learning (ML) research and practice. So far, however, little is known about the common practice as well as bottlenecks faced by the community in tackling the research questions posed. To shed light on the status quo of algorithm development in the specific field of biomedical imaging analysis, we designed an international survey that was issued to all participants of challenges conducted in conjunction with the IEEE ISBI 2021 and MICCAI 2021 conferences (80 competitions in total). The survey covered participants' expertise and working environments, their chosen strategies, as well as algorithm characteristics. A median of 72% challenge participants took part in the survey. According to our results, knowledge exchange was the primary incentive (70%) for participation, while the reception of prize money played only a minor role (16%). While a median of 80 working hours was spent on method development, a large portion of participants stated that they did not have enough time for method development (32%). 25% perceived the infrastructure to be a bottleneck. Overall, 94% of all solutions were deep learning-based. Of these, 84% were based on standard architectures. 43% of the respondents reported that the data samples (e.g., images) were too large to be processed at once. This was most commonly addressed by patch-based training (69%), downsampling (37%), and solving 3D analysis tasks as a series of 2D tasks. K-fold cross-validation on the training set was performed by only 37% of the participants and only 50% of the participants performed ensembling based on multiple identical models (61%) or heterogeneous models (39%). 48% of the respondents applied postprocessing steps.

We present the Verifee Dataset: a novel dataset of news articles with fine-grained trustworthiness annotations. We develop a detailed methodology that assesses the texts based on their parameters encompassing editorial transparency, journalist conventions, and objective reporting while penalizing manipulative techniques. We bring aboard a diverse set of researchers from social, media, and computer sciences to overcome barriers and limited framing of this interdisciplinary problem. We collect over $10,000$ unique articles from almost $60$ Czech online news sources. These are categorized into one of the $4$ classes across the credibility spectrum we propose, raging from entirely trustworthy articles all the way to the manipulative ones. We produce detailed statistics and study trends emerging throughout the set. Lastly, we fine-tune multiple popular sequence-to-sequence language models using our dataset on the trustworthiness classification task and report the best testing F-1 score of $0.52$. We open-source the dataset, annotation methodology, and annotators' instructions in full length at https://verifee.ai/research to enable easy build-up work. We believe similar methods can help prevent disinformation and educate in the realm of media literacy.

In this paper, we present a modified Xception architecture, the NEXcepTion network. Our network has significantly better performance than the original Xception, achieving top-1 accuracy of 81.5% on the ImageNet validation dataset (an improvement of 2.5%) as well as a 28% higher throughput. Another variant of our model, NEXcepTion-TP, reaches 81.8% top-1 accuracy, similar to ConvNeXt (82.1%), while having a 27% higher throughput. Our model is the result of applying improved training procedures and new design decisions combined with an application of Neural Architecture Search (NAS) on a smaller dataset. These findings call for revisiting older architectures and reassessing their potential when combined with the latest enhancements.

Vision Transformers convert images to sequences by slicing them into patches. The size of these patches controls a speed/accuracy tradeoff, with smaller patches leading to higher accuracy at greater computational cost, but changing the patch size typically requires retraining the model. In this paper, we demonstrate that simply randomizing the patch size at training time leads to a single set of weights that performs well across a wide range of patch sizes, making it possible to tailor the model to different compute budgets at deployment time. We extensively evaluate the resulting model, which we call FlexiViT, on a wide range of tasks, including classification, image-text retrieval, open-world detection, panoptic segmentation, and semantic segmentation, concluding that it usually matches, and sometimes outperforms, standard ViT models trained at a single patch size in an otherwise identical setup. Hence, FlexiViT training is a simple drop-in improvement for ViT that makes it easy to add compute-adaptive capabilities to most models relying on a ViT backbone architecture. Code and pre-trained models are available at https://github.com/google-research/big_vision

Warning: this paper contains content that may be offensive or upsetting. Considering the large amount of content created online by the minute, slang-aware automatic tools are critically needed to promote social good, and assist policymakers and moderators in restricting the spread of offensive language, abuse, and hate speech. Despite the success of large language models and the spontaneous emergence of slang dictionaries, it is unclear how far their combination goes in terms of slang understanding for downstream social good tasks. In this paper, we provide a framework to study different combinations of representation learning models and knowledge resources for a variety of downstream tasks that rely on slang understanding. Our experiments show the superiority of models that have been pre-trained on social media data, while the impact of dictionaries is positive only for static word embeddings. Our error analysis identifies core challenges for slang representation learning, including out-of-vocabulary words, polysemy, variance, and annotation disagreements, which can be traced to characteristics of slang as a quickly evolving and highly subjective language.

Warning: this paper contains content that may be offensive or upsetting. In the current context where online platforms have been effectively weaponized in a variety of geo-political events and social issues, Internet memes make fair content moderation at scale even more difficult. Existing work on meme classification and tracking has focused on black-box methods that do not explicitly consider the semantics of the memes or the context of their creation. In this paper, we pursue a modular and explainable architecture for Internet meme understanding. We design and implement multimodal classification methods that perform example- and prototype-based reasoning over training cases, while leveraging both textual and visual SOTA models to represent the individual cases. We study the relevance of our modular and explainable models in detecting harmful memes on two existing tasks: Hate Speech Detection and Misogyny Classification. We compare the performance between example- and prototype-based methods, and between text, vision, and multimodal models, across different categories of harmfulness (e.g., stereotype and objectification). We devise a user-friendly interface that facilitates the comparative analysis of examples retrieved by all of our models for any given meme, informing the community about the strengths and limitations of these explainable methods.