---
title: 'MICACL: Category-Aware Contrastive Learning for DFER'
url: https://www.emergentmind.com/topics/micacl
type: topic
---

# MICACL: Category-Aware Contrastive Learning for DFER

Searching arXiv for MICACL/MICAL to verify naming and citation targets.
MICACL most specifically denotes **Multi-Instance Category-Aware Contrastive Learning**, a framework for **long-tailed dynamic facial expression recognition** that combines multi-instance learning, adaptive spatio-temporal interaction modeling, importance-weighted temporal aggregation, and category-aware contrastive optimization [2509.04344]. In the arXiv literature, however, the string **“MICACL”** is not uniformly used: one seizure-detection paper states that **MICACL** is a misspelling of **MICAL**—“Mutual Information-based CNN-Aided Learned factor graphs” [2206.02298]—and one speech-based Mild Cognitive Impairment study uses **MICACL** as a shorthand label in a synthesis of acoustic classification [2408.16732]. The dominant explicit acronym expansion attached to **MICACL** itself is therefore the long-tailed dynamic facial expression recognition method introduced in [2509.04344].

## 1. Nomenclature and disambiguation

The acronym has multiple nearby usages in recent preprints, but they refer to different research problems and method families.

| String | Expansion or usage | Domain |
|---|---|---|
| MICACL | Multi-Instance Category-Aware Contrastive Learning | Long-tailed dynamic facial expression recognition |
| MICAL | Mutual Information-based CNN-Aided Learned factor graphs | Seizure detection from EEG |
| MICACL | Shorthand label in a synthesis of Mild Cognitive Impairment acoustic classification | Speech-based cognitive assessment |

The seizure-detection work explicitly states that there is **no alternative acronym “MICACL”** for its method and that the appearance of that form is a **misspelling of MICAL** [2206.02298]. By contrast, the dynamic facial expression recognition paper uses **MICACL** as the formal name of the method and expands it as **Multi-Instance Category-Aware Contrastive Learning** [2509.04344].

A plausible implication is that references to **MICACL** require contextual disambiguation. In computer vision, it refers to the DFER framework of [2509.04344]; in EEG seizure detection, the correct acronym is **MICAL** rather than **MICACL** [2206.02298].

## 2. Problem setting in dynamic facial expression recognition

MICACL addresses **dynamic facial expression recognition (DFER) in the wild**, where two coupled difficulties are emphasized: **long-tailed category distributions** and **complex spatio-temporal feature modeling** [2509.04344]. The paper frames each video \(V\) as a **bag** \(B\) of \(T\) instances under a standard **multi-instance learning** formulation. In this work, **\(T = 16\) frames per video are sampled for training and inference**, and the bag carries a **video-level emotion label** while **instance labels are not available** [2509.04344].

The representation is built on a **standard 3D CNN backbone (R3D from Torchvision) with pretrained weights**, with videos resized to **\(112 \times 112\)**. The backbone produces per-time features that are spatially pooled into instance embeddings, written in the paper as
\[
\mathbf{X} \in \mathbb{R}^{B \times T \times C},
\]
with batch size \(B\), \(T\) instances per bag, and channel dimension \(C\) [2509.04344].

The classification target is optimized with a **bag-level cross-entropy loss** over \(C\) classes:
\[
\mathcal{L}_{\mathrm{CE}} = - \sum_{c=1}^C \mathbb{1}[y=c]\log p_c.
\]
The paper states that an auxiliary **Multiscale Category-aware Contrastive Learning (MCCL)** loss is added to counteract long-tailed bias [2509.04344].

The motivating claim is that head classes occupy a larger learned feature space, while minority classes suffer from biased decision boundaries. This suggests that MICACL is designed not merely as a temporal recognizer, but as a rebalancing mechanism for supervised video representation learning under heavy class skew.

## 3. Core architecture: GEIIM, WIAN, and MCCL

MICACL is organized around three named components: the **Graph-Enhanced Instance Interaction Module (GEIIM)**, the **Weighted Instance Aggregation Network (WIAN)**, and **Multiscale Category-aware Contrastive Learning (MCCL)** [2509.04344].

### GEIIM

GEIIM models **inter-instance dependencies** through a learned adjacency matrix. The paper defines node embeddings \(\mathbf{N}_1 \in \mathbb{R}^{T \times d}\) and \(\mathbf{N}_2 \in \mathbb{R}^{d \times T}\), and computes the adaptive adjacency as
\[
\mathbf{A} = \mathrm{Softmax}\big(\mathrm{ReLU}(\mathbf{N}_1 \cdot \mathbf{N}_2)\big).
\]
The stated rationale is that **ReLU suppresses low-quality/noisy relations** and **Softmax normalizes affinities to stabilize relation modeling** [2509.04344].

Instead of a normalized GCN update, MICACL uses a **linear diffusion-style interaction**
\[
\mathbf{H} = \alpha \mathbf{X} + (1 - \alpha)\,\mathbf{A}\cdot \mathbf{X}, \quad \alpha \in [0,1],
\]
with learnable \(\alpha\) balancing direct instance information and neighborhood-aggregated information [2509.04344]. GEIIM then applies **multiscale temporal convolutions** to capture local-to-global temporal patterns, although the paper does not numerically specify kernel sizes or dilation rates.

### WIAN

WIAN aggregates instances into a bag representation while emphasizing salient frames and attenuating noise. It augments an LSTM with a **Dynamic Weight Gate (DWG)** driven by instance importance. The importance weights are computed from the GEIIM outputs as
\[
w_t = \mathrm{Softmax}(\mathbf{H}_t,\ \mathrm{dim}=-1).
\]
The gate is then defined as
\[
\mathbf{d}_t = \sigma\big(\mathbf{W}_s \mathbf{x}_t + \mathbf{U}_s \mathbf{h}_{t-1}\big)\cdot w_t.
\]
The modified LSTM cell update is
\[
\mathbf{c}_t = \mathbf{f}_t \cdot \mathbf{c}_{t-1} + \mathbf{i}_t \cdot \tilde{\mathbf{c}}_t \cdot \mathbf{d}_t,
\qquad
\mathbf{h}_t = \mathbf{o}_t \cdot \tanh(\mathbf{c}_t).
\]
After recurrent aggregation, WIAN applies **multi-head self-attention** to \(\{\mathbf{h}_t\}_{t=1}^T\) to produce a bag-level representation for classification [2509.04344].

### MCCL

MCCL is the long-tail mitigation component. It defines a **category-aware weight**
\[
\mathbf{W}_\mathrm{c}(i) = \frac{1}{n_{\mathrm{c}[i]}\cdot \big(1 - \mathrm{softmax}(\mathbf{y}_\mathrm{p})[i]\big)},
\]
where \(n_{\mathrm{c}[i]}\) is the sample count of category \(i\), and \(\mathbf{y}_\mathrm{p}\) denotes predicted class probabilities [2509.04344]. This weighting is explicitly described as inversely weighting frequent classes and upweighting hard-to-classify samples.

MCCL builds multiscale features through
\[
\mathbf{X}'_s = \mathrm{Proj}_s\big(\mathrm{Pool}_s(\mathbf{X}')\big),
\]
computes cosine similarity with temperature \(\tau\),
\[
\mathbf{S}_{ij}^s = \frac{\mathbf{X}'_s[i]\cdot \mathbf{X}'_s[j]}{\tau\,\|\mathbf{X}'_s[i]\|\,\|\mathbf{X}'_s[j]\|},
\]
and adjusts the temperature using the dataset class distribution [2509.04344]. The scale-specific contrastive objective is
\[
\mathcal{L}_\mathrm{c}^s = -\frac{1}{B} \sum_{i=1}^{B} \frac{\sum_{j\in \mathcal{P}(i)} \mathbf{W}_\mathrm{c}(i)\cdot \exp(\mathbf{S}_{ij}^s)}{\sum_{k=1}^{B} \exp(\mathbf{S}_{ik}^s)},
\]
with multiscale aggregation
\[
\mathcal{L}_\mathrm{MC} = \frac{1}{|S|} \sum_{s\in S} \mathcal{L}_\mathrm{c}^s,
\qquad
\mathcal{L}_\mathrm{all} = \mathcal{L}_\mathrm{MC} + \mathcal{L}_\mathrm{CET}.
\]

The paper further states that MCCL includes a **Classification Enhancement Term (CET)** combining feature augmentation regularization with an intensity-aware loss, with details placed in the appendix [2509.04344].

## 4. Training configuration and computational profile

The training pipeline uses **R3D (Torchvision pretrained)** as the backbone, **AdamW** as the optimizer, and a **cosine learning rate** schedule over **300 epochs**, with **initial learning rate \(4\mathrm{e}{-4}\)**, **minimum learning rate \(3\mathrm{e}{-6}\)**, and **weight decay \(0.05\)** [2509.04344]. The reported hardware is an **RTX A6000**.

The paper does not report an explicit resampling strategy or memory bank. Instead, class balancing is handled through the **MCCL weighting** mechanism and the multiscale contrastive objective [2509.04344]. The hidden dimension \(d\) in GEIIM, the value of \(\alpha\), the multiscale convolution settings, and the coefficient that would weight MCCL relative to cross-entropy are not specified in the reported summary.

The computational footprint is reported as **1.69 G FLOPs** in the DFEW table [2509.04344]. Parameter counts and runtime are not reported.

A plausible implication is that MICACL was positioned as a comparatively efficient long-tailed DFER architecture: it retains an R3D backbone rather than replacing it with a heavier temporal transformer, while moving much of the inductive bias into graph interaction, gated aggregation, and contrastive reweighting.

## 5. Empirical performance on DFEW and FERV39K

MICACL is evaluated on **DFEW** and **FERV39K**, using **WAR** and **UAR** as principal metrics [2509.04344]. DFEW is described as containing **16,000+ clips from 1,500+ films**, with **7 basic expressions** and **5-fold cross-validation**. FERV39K contains **38,935 clips across 22 sub-scenes** [2509.04344].

### Main results

| Dataset | MICACL performance | Comparison noted in the paper |
|---|---|---|
| DFEW | WAR 69.56, UAR 63.21 | Best WAR and UAR simultaneously |
| FERV39K | WAR 48.57, UAR 40.25 | Highest WAR and UAR |

On **DFEW**, MICACL reports per-emotion accuracies of **Happiness 89.22**, **Sadness 69.48**, **Neutral 63.72**, **Anger 73.47**, **Surprise 60.92**, **Disgust 29.65**, and **Fear 42.57**, together with **WAR 69.56**, **UAR 63.21**, and **FLOPs 1.69 G** [2509.04344]. The same table reports the following selected comparisons: **NR-DFERNet** with **WAR 68.19**, **UAR 54.21**, **Disgust 0.00**, **Fear 19.43**; **Former-DFER** with **WAR 65.70**, **UAR 53.69**, **Disgust 3.45**, **Fear 31.78**; **STT** with **WAR 66.45**, **UAR 54.58**, **Disgust 3.49**, **Fear 34.04**; and **M3DFEL** with **WAR 69.25**, **UAR 56.10**, **Disgust 0.00**, **Fear 31.63** [2509.04344].

On **FERV39K**, MICACL reports **WAR 48.57** and **UAR 40.25**, compared with **LOGO-Former** at **WAR 48.13**, **UAR 38.22**; **Former-DFER** at **WAR 46.85**, **UAR 37.20**; and **M3DFEL** at **WAR 47.67**, **UAR 35.94** [2509.04344].

The paper emphasizes the tail-class gains on DFEW, especially **Disgust 29.65%** and **Fear 42.57%**, in contrast to prior works that report **near-zero Disgust** and lower Fear performance [2509.04344]. This supports the intended function of MCCL as a mechanism for reducing head-class domination in the learned representation.

## 6. Ablation findings, interpretation, and limitations

The ablation study on **DFEW fold 5** reports the following **WAR/UAR** progression: **Baseline (R3D + Self-Attention + Conv1D; CE loss)** at **64.55 / 54.17**; **+ GEIIM** at **68.11 / 57.39**; **+ GEIIM + WIAN** at **68.62 / 56.26**; and **+ GEIIM + WIAN + MCCL** at **69.91 / 64.34** [2509.04344]. The accompanying interpretation is explicit: **GEIIM substantially boosts both WAR and UAR**, **WIAN contributes additional gains, especially in WAR**, and **MCCL delivers the largest UAR improvement (+8.08 over baseline)**.

The paper also notes that a **t-SNE visualization** shows **tighter intra-class clusters and clearer inter-class separation** than the baseline, which is presented as consistent with the combined effect of MCCL and the GEIIM+WIAN aggregation pipeline [2509.04344]. This suggests that MICACL is meant to alter both optimization geometry and temporal evidence accumulation, rather than only reweighting the final classification loss.

The reported limitations are concrete. The adjacency matrix \(\mathbf{A}\) is learned from **node embeddings** rather than directly from content similarity between instance features; **multiscale convolution settings are not specified**; **MCCL relies on class counts and predicted probabilities**, so inaccurate early predictions may destabilize weighting; and **R3D may be less expressive than temporal transformers**, although transformers incur higher compute cost [2509.04344]. Proposed extensions include **content-adaptive graph learning**, **prototype-based supervised contrastive learning with a memory bank**, **effective-number class weighting**, **temporal transformers or hybrid CNN-Transformer backbones**, and **dynamic sampling strategies for long-tailed distributions** [2509.04344].

In relation to neighboring uses of the acronym, the seizure-detection method **MICAL** offers a useful contrast: it combines **neural mutual information estimation**, a **1D CNN**, and **factor graph inference** for multichannel EEG seizure detection on CHB-MIT [2206.02298]. The Mild Cognitive Impairment speech paper, by contrast, studies **automatic detection of Mild Cognitive Impairment using high-dimensional acoustic features in spontaneous speech**, comparing **Random Forests**, **Sparse Logistic Regression**, **k-Nearest Neighbors**, **Sparse Support Vector Machine**, and **Decision Tree** on the TAUKADIAL dataset [2408.16732]. These works are methodologically distinct from MICACL proper, and their juxtaposition mainly underscores that the string **MICACL** is ambiguous unless grounded by domain context and citation.

## 7. Position within current research usage

Within the supplied arXiv record, **MICACL** is best understood as a **multi-instance, long-tail-aware DFER framework** whose distinguishing combination is: **adaptive inter-instance graph modeling** through GEIIM, **importance-aware recurrent aggregation** through WIAN, and **multiscale category-aware contrastive optimization** through MCCL [2509.04344]. Its reported empirical profile is defined by simultaneous gains in **WAR** and **UAR**, especially on minority categories such as **Disgust** and **Fear**, and by an ablation pattern in which the largest balanced-performance improvement is attributed to MCCL [2509.04344].

The broader terminological landscape remains unsettled. One paper treats **MICACL** as a **misspelling** of **MICAL** in EEG seizure detection [2206.02298], while another uses the string as a label associated with **Mild Cognitive Impairment acoustic classification** rather than as a named method [2408.16732]. In encyclopedia usage, the most precise treatment is therefore to reserve **MICACL** for **Multi-Instance Category-Aware Contrastive Learning** unless the surrounding literature clearly indicates one of the alternative contexts.

Source: https://www.emergentmind.com/topics/micacl