In this work, we propose a novel supervised contrastive loss that enables the
integration of taxonomic hierarchy information during the representation
learning process. A supervised contrastive loss operates by enforcing that
images with the same class label (positive samples) project closer to each
other than images with differing class labels (negative samples). The advantage
of this approach is that it directly penalizes the structure of the
representation space itself. This enables greater flexibility with respect to
encoding semantic concepts. However, the standard supervised contrastive loss
only enforces semantic structure based on the downstream task (i.e. the class
label). In reality, the class label is only one level of a \emph{hierarchy of
different semantic relationships known as a taxonomy}. For example, the class
label is oftentimes the species of an animal, but between different classes
there are higher order relationships such as all animals with wings being
``birds". We show that by explicitly accounting for these relationships with a
weighting penalty in the contrastive loss we can out-perform the supervised
contrastive loss. Additionally, we demonstrate the adaptability of the notion
of a taxonomy by integrating our loss into medical and noise-based settings
that show performance improvements by as much as 7%.

我们提出了一种新型的监督对比损失函数，在表示学习过程中引入了分类树结构信息。该损失函数通过强制要求具有相同类标签（正样本）的图像在投影空间中比具有不同类标签（负样本）的图像更接近彼此来实现。该方法的优势在于它直接惩罚表示空间的结构，这使得在编码语义概念方面具有更大的灵活性。然而，标准的监督对比损失函数只基于下游任务（即类标签）强制执行语义结构。我们展示了通过在对比损失中明确考虑这些关系，采用加权惩罚的方式，可以超过监督对比损失。此外，我们将分类树的概念融入医学和噪声环境的设置中，表现出高达 7% 的性能提升。

税收是您所需的一切：将分类层次关系集成到对比损失中

Taxes Are All You Need: Integration of Taxonomical Hierarchy  Relationships into the Contrastive Loss

We focus on the problem of species distribution modeling using global-scale
presence-only data. Most previous studies have mapped the range of a given
species using geographical and environmental features alone. To capture a
stronger implicit relationship between species, we encode the taxonomic
hierarchy of species using a large language model. This enables range mapping
for any taxonomic rank and unseen species without additional supervision.
Further, we propose a novel proximity-aware evaluation metric that enables
evaluating species distribution models using any pixel-level representation of
ground-truth species range map. The proposed metric penalizes the predictions
of a model based on its proximity to the ground truth. We describe the
effectiveness of our model by systematically evaluating on the task of species
range prediction, zero-shot prediction and geo-feature regression against the
state-of-the-art. Results show our model outperforms the strong baselines when
trained with a variety of multi-label learning losses.

利用全球范围的仅有存在数据，我们解决物种分布建模问题。通过使用大规模语言模型对物种的分类学阶层进行编码，我们能够捕捉物种之间更强的隐式关系，并对任意分类水平和未知物种进行范围映射，而不需要额外监督。此外，我们提出了一种新颖的基于接近度的评估度量，能够使用任何像素级物种分布图的表示方式来评估物种分布模型。所提出的度量标准惩罚模型根据其与实际情况的接近程度对预测结果进行评估。通过在物种范围预测、零 - shot 预测和地理特征回归等任务上进行系统评估，我们论述了我们模型的有效性。结果显示，我们的模型在使用各种多标签学习损失进行训练时优于强基准模型。

LD-SDM: 语言驱动的层次物种分布建模

LD-SDM: Language-Driven Hierarchical Species Distribution Modeling

Commonsense knowledge (CSK) supports a variety of AI applications, from
visual understanding to chatbots. Prior works on acquiring CSK, such as
ConceptNet, have compiled statements that associate concepts, like everyday
objects or activities, with properties that hold for most or some instances of
the concept. Each concept is treated in isolation from other concepts, and the
only quantitative measure (or ranking) of properties is a confidence score that
the statement is valid. This paper aims to overcome these limitations by
introducing a multi-faceted model of CSK statements and methods for joint
reasoning over sets of inter-related statements. Our model captures four
different dimensions of CSK statements: plausibility, typicality, remarkability
and salience, with scoring and ranking along each dimension. For example,
hyenas drinking water is typical but not salient, whereas hyenas eating
carcasses is salient. For reasoning and ranking, we develop a method with soft
constraints, to couple the inference over concepts that are related in in a
taxonomic hierarchy. The reasoning is cast into an integer linear programming
(ILP), and we leverage the theory of reduction costs of a relaxed LP to compute
informative rankings. This methodology is applied to several large CSK
collections. Our evaluation shows that we can consolidate these inputs into
much cleaner and more expressive knowledge. Results are available at
this https URL

本文提出了一种多维模型的常识知识语句，以及一种关于相关语句集合的联合推理方法。通过将推理转化为整数线性规划，采用软约束和松弛 LP 的降维成本理论来计算信息化排序，本方法可以应用于多个大型 CSK 收集，最终可以将这些内容转化为更干净和更具表现力的知识。