---
title: Compositional Zero-shot Learning (CZSL)
url: https://www.emergentmind.com/topics/compositional-zero-shot-learning-czsl-7c83b33f-fcdf-41be-90cc-5b5dac01ed6d
type: topic
---

# Compositional Zero-shot Learning (CZSL)

Compositional Zero-shot Learning (CZSL) is a paradigm in visual recognition aiming to identify novel compositions of semantic primitives—typically attributes (or states) and objects—unseen during training. Unlike standard zero-shot learning, which treats classes as atomic, CZSL leverages the combinatorial structure of attributes and objects, allowing for the transfer of learned concepts to new, unobserved pairs. The core challenge is that the appearance of primitives is contextual; for example, "small plane" and "small cat" vary in visual realization depending on the object context. Robust CZSL requires modeling contextual entanglement, generalizing to combinatorially many pairs, and maintaining discriminability against long-tail data distributions.

## 1. Problem Formulation and Challenges

CZSL operates over a label space $\mathcal{C} = \mathcal{A} \times \mathcal{O}$, partitioned into seen $(\mathcal{C}_{train})$ and unseen $(\mathcal{C}_{unseen})$ compositions. Images $\mathbf{x}$ in the training set are labeled only with seen pairs, while the evaluation protocols include closed-world CZSL (predict among $\mathcal{C}_{train} \cup \mathcal{C}_{unseen}$) and open-world CZSL (predict among all possible $\mathcal{A} \times \mathcal{O}$, including infeasible pairs) [2510.11106].

The main challenges are:

- **Contextual entanglement:** Primitives are not visually independent. The appearance of an attribute is modulated by the associated object, producing a highly contextual recognition problem [2507.17377].
- **Combinatorial explosion:** With $n$ attributes and $m$ objects, there exist $nm$ possible pairs, but only a sparse subset is observed during training [2510.11106].
- **Long-tailed data:** Natural distributions over compositions tend to be highly imbalanced, with many rare or minority combinations [2312.15923].
- **Open world distractors:** In open-world CZSL, the label space contains many implausible or visually unsupported pairs ("hairy apple"), creating significant challenges for recognition algorithms [2101.12609, 2105.01017].

## 2. CZSL Methodological Taxonomy: Disentanglement Strategies

Munir et al. [2510.11106] introduce a taxonomy grounded in the principle of disentanglement, categorizing CZSL methods by the modality and level where primitive separation occurs:

- **No explicit disentanglement:** Treat attribute-object pairs as atomic, fuse primitive tokens directly in text, or use joint composition graphs without primitive separation. Examples: RedWine (linear composition), Compositional Soft Prompting (CSP), CompCos, Co-CGE [2105.01017].
- **Textual disentanglement:** Refine primitive word embeddings and learn composition in text space, e.g., AoP (attribute as operator), DFSP (separated soft prompts) [2211.10681]. These approaches do not model context-dependent visual variability.
- **Visual disentanglement:** Extract dedicated "attribute" and "object" features from visual encoders using attention, prototype networks, or auxiliary losses; align each with textual prototypes. Methods include ADE (cross-attention disentangler) [2303.15111], CANet (conditional attributes) [2305.17940], CDS-CZSL (context- and diversity-weighted attributes) [2402.17251], and HOPE (Hopfield memory + mixture of experts) [2311.14747].
- **Cross-modal (hybrid) disentanglement:** Decompose primitives in both language and vision, then align and jointly optimize them in a shared embedding space. Representative methods: PLID (language-informed class distributions and visual-language primitive decomposition) [2305.14428], LPR (probabilistic relations via cross-attention) [2501.14308], CAMS (gated cross-attention, multi-space disentanglement) [2511.16378], and TRIDENT (MLLM embeddings, attribute smoothing) [2411.12584].

## 3. Model Architectures and Algorithmic Innovations

CZSL architectures coalesce around several key mechanisms:

- **Prompt engineering in VLMs:** Tune soft/hard prompts for primitive concepts, sometimes using LLM-generated class distributions (PLID). Fusing separate prompts (e.g., pair + state + object) and decomposing their contributions (DFSP) boosts discrimination and generalization [2211.10681, 2501.17171].
- **Attention-based disentanglers:** Employ cross-attention or multi-head self-attention to extract primitive-exclusive features (ADE, CAMS, TRIDENT) [2303.15111, 2511.16378, 2411.12584].
- **Probabilistic and conditional modeling:** Decompose $p(o, a \mid x) = p(o \mid x) p(a \mid o, x)$ to capture compositional dependencies (CPF) [2507.17377]. CANet conditions attribute embeddings on both object and image context [2305.17940].
- **Graph-based embedding propagation:** Use GCNs over primitive and composition nodes, propagating information from seen to unseen compositions (Co-CGE). Feasibility scores weight graph edges and impose margin penalties [2105.01017].
- **Feature augmentation and debiasing:** Synthesize novel feature representations from disentangled subspaces to support generalization to rare or unseen pairs; debiasing weights are computed from training statistics (DeFA) [2509.12711, 2312.15923].
- **Mixture of experts and memory networks:** Retrieve relevant prototypes via Modern Hopfield Networks and combine them using soft mixture of experts (HOPE), enhancing generalization to unseen compositions [2311.14747].
- **Contrastive and regularization objectives:** Employ label smoothing via auxiliary LLM-generated attributes (TRIDENT), adaptive contrastive loss with hard negatives (ULAO), and orthogonal/EMD-based regularizers (ADE) [2411.12584, 2412.07161, 2303.15111].
- **Prompt-guided fusion and inter/intra-modal fusion:** Hierarchical fusion modules merge vision-language streams (separated inter/intra-modal fusion prompts) [2501.17171].

## 4. Empirical Protocols, Benchmarks, and Evaluation Metrics

CZSL is assessed under closed-world and open-world conditions using three principal benchmarks:

**Datasets**

| Dataset     | Attributes | Objects | Seen pairs | Unseen pairs | Images    |
|-------------|------------|---------|------------|--------------|-----------|
| MIT-States  | 115        | 245     | 1262       | 700+         | 53k       |
| UT-Zappos   | 16         | 12      | 83         | 33           | 50k       |
| C-GQA       | 413        | 674     | 5592       | 3932         | 27k       |

**Metrics**

- **Seen accuracy (S) and unseen accuracy (U):** Top-1 accuracy on seen and unseen pairs.
- **Harmonic mean (HM):** $HM = 2SU / (S+U)$ penalizes tradeoffs between seen and unseen.
- **Area Under Curve (AUC):** Measures the area under the seen–unseen curve as calibration bias sweeps.
- **Primitive accuracies:** Performance of separate attribute and object classifiers.

**Results**
Recent models achieve state-of-the-art HM and AUC under both protocols:

| Method      | MIT-States HM | UT-Zappos HM | C-GQA HM | MIT-States AUC | UT-Zappos AUC | C-GQA AUC |
|-------------|---------------|--------------|----------|----------------|---------------|-----------|
| DFSP        | 37.3          | 47.2         | 27.1     | 20.6           | 36.0          | 10.5      |
| PLID        | 39.0          | 52.4         | 27.9     | 22.1           | 38.7          | 11.0      |
| DeFA        | 39.3          | 58.6         | 32.3     | 22.8           | 46.1          | 14.6      |
| CAMS        | 41.0          | 58.5         | 36.4     | 24.2           | 47.4          | 17.4      |
| TRIDENT     | 30.9          | 23.4 (VAW)   | 22.6     | 14.2           | 8.3 (VAW)     | 8.0       |

This table summarizes HM and AUC (closed-world) for recent CZSL models [2509.12711, 2501.14308, 2411.12584, 2511.16378].

## 5. Contextuality, Specificity, and Long-Tail Phenomena

Recent work emphasizes the need to model:

- **Contextuality:** Attribute manifestation varies dramatically with object, requiring conditional representations or object-guided attention (CPF, CANet, CDS-CZSL) [2507.17377, 2305.17940, 2402.17251].
- **Specificity:** Some attributes are highly informative for certain objects ("Sliced-Strawberry" vs "Red-Strawberry") [2402.17251]. Specificity is measured via object diversity and context, and is used to refine attribute scores and to prune large open-world search spaces.
- **Class imbalance / long-tail:** The distribution over compositions is highly skewed in real data, and visual bias can effectively under-represent certain classes. Estimation and integration of proximate class priors (ProLT) or debiased augmentation weights are used to achieve more balanced predictions [2312.15923, 2509.12711].

## 6. Robustness, Open-World Generalization, and Future Directions

CZSL methods face pronounced AUC drops in open-world settings, where models must contend with many implausible or unsupported pairs. Feasibility scores (CompCos, Co-CGE), graph-based propagation, conditioned attribute generation, and compositional mixup are among the approaches enhancing open-world robustness [2101.12609, 2105.01017, 2208.10046].

Emerging lines—such as multimodal large language models (MLLMs), cross-modal fusion, hierarchical decomposition, and continual adaptation—address open issues: modeling fine-grained contextuality, handling novel primitives, integrating multi-attribute, multi-object, or hierarchical labels, and achieving parameter-efficient adaptation for large VLM backbones [2510.11106, 2411.12584, 2511.16378].

Persistent challenges include:

- **Modeling complex attribute–object interactions.**
- **Generalizing to compositional primitives with no prior training examples.**
- **Intrinsic robustness to open-world distractors beyond feasibility filtering.**
- **Efficient integration of knowledge graphs, external ontologies, and large-scale MLLMs.**

CZSL is increasingly supported by unified taxonomies and comparative analyses [2510.11106], and by the release of large-scale benchmarks under reference-limited and open-world protocols [2208.10046].

## 7. Research Impact and Theoretical Insights

The field is informed by both theoretical and neuroscientific insights. For example, DeFA draws inspiration from studies of visual imagination, synthesizing features for unseen compositions via a disentangle-and-reconstruct pipeline [2509.12711]. Probabilistic and conditional decompositions (CPF, CANet, LPR) and graph-based propagation (Co-CGE, MetaCGL) provide principled solutions for contextuality and transfer.

The survey by Munir et al. [2510.11106] highlights that deeper integration of vision and language—via fine-grained disentanglement, hybrid architectures, and compositional prompting—in the context of large-scale MLLMs, offers a promising trajectory for future developments.

---

## Table: Representative CZSL Algorithms by Disentanglement Family

| Family                       | Key Approaches                                | Typical Mechanism                         |
|------------------------------|-----------------------------------------------|-------------------------------------------|
| No Disentanglement           | RedWine, CSP, CompCos, Co-CGE                | Joint composition modeling, feasibility   |
| Textual Disentanglement      | AoP, DFSP, BMP-Net, ASP                      | Soft/hard prompts, operator-based fusion  |
| Visual Disentanglement       | ADE, CANet, CDS-CZSL, HOPE, CAMS             | Cross-attention, conditional attributes   |
| Cross-Modal (Hybrid)         | PLID, LPR, CAMS, TRIDENT, CAILA              | Joint vision-language decomposition       |

---

Recent advances in CZSL demonstrate the necessity of explicit primitive disentanglement, contextual modeling, compositional augmentation, and hybrid fusion for state-of-the-art zero-shot recognition. Ongoing work continues to extend CZSL to more realistic open-world regimes and richer compositional structures.

Source: https://www.emergentmind.com/topics/compositional-zero-shot-learning-czsl-7c83b33f-fcdf-41be-90cc-5b5dac01ed6d