目次
The classification conundrum
The R.E.D. algorithm
How it works
Greedy subset selection with least similar elements
Semi-supervised classification with noise oversampling
Proxy active learning via an LLM agent
ホームページ テクノロジー周辺機器 AI R.E.D。:専門家の代表団によるテキスト分類のスケーリング

R.E.D。:専門家の代表団によるテキスト分類のスケーリング

Mar 22, 2025 am 10:33 AM

With the new age of problem-solving augmented by Large Language Models (LLMs), only a handful of problems remain that have subpar solutions. Most classification problems (at a PoC level) can be solved by leveraging LLMs at 70–90% Precision/F1 with just good prompt engineering techniques, as well as adaptive in-context-learning (ICL) examples.

What happens when you want to consistently achieve performancehigher than that — when prompt engineering no longer suffices?

The classification conundrum

Text classification is one of the oldest and most well-understood examples of supervised learning. Given this premise, it shouldreallynot be hard to build robust, well-performing classifiers that handle a large number of input classes, right…?

Welp. It is.

It actually has to do a lot more with the ‘constraints’ that the algorithm is generally expected to work under:

  • low amount of training data per class
  • high classification accuracy (that plummets as you add more classes)
  • possible addition ofnew classesto an existing subset of classes
  • quick training/inference
  • cost-effectiveness
  • (potentially) really large number of training classes
  • (potentially) endlessrequiredretraining ofsomeclasses due to data drift, etc.

Ever tried building a classifier beyond a few dozen classes under these conditions? (I mean, even GPT could probably do a great job up to ~30 text classes with just a few samples…)

Considering you take the GPT route — If you have more than a couple dozen classes or a sizeable amount of data to be classified, you are gonna have to reach deep into your pockets with the system prompt, user prompt, few shot example tokens that you will need to classifyone sample.That is after making peace with the throughput of the API, even if you are running async queries.

In applied ML, problems like these are generally tricky to solve since they don’t fully satisfy the requirements of supervised learning or aren’t cheap/fast enough to be run via an LLM. This particular pain point is what the R.E.D algorithm addresses: semi-supervised learning, when the training data per class is not enough to build (quasi)traditional classifiers.

The R.E.D. algorithm

R.E.D: Recursive Expert Delegationis a novel framework that changes how we approach text classification. This is an applied ML paradigm — i.e., there is nofundamentally differentarchitecture to what exists, but its a highlight reel of ideas that work best to build something that is practical and scalable.

In this post, we will be working through a specific example where we have a large number of text classes (100–1000), each class only has few samples (30–100), and there are a non-trivial number of samples to classify (10,000–100,000). We approach this as asemi-supervised learningproblem via R.E.D.

Let’s dive in.

How it works

R.E.D。:専門家の代表団によるテキスト分類のスケーリング

Instead of having a single classifier classify between a large number of classes, R.E.D. intelligently:

  1. Divides and conquers— Break the label space (large number of input labels) into multiple subsets of labels. This is a greedy label subset formation approach.
  2. Learns efficiently— Trains specialized classifiers for each subset. This step focuses on building a classifier that oversamples on noise, where noise is intelligently modeled as data fromother subsets.
  3. Delegates to an expert— Employes LLMs as expert oracles for specific label validation and correction only, similar to having a team of domain experts. Using an LLM as a proxy, it empirically ‘mimics’howa human expert validates an output.
  4. Recursive retraining— Continuously retrains with fresh samples added back from the expert until there are no more samples to be added/a saturation from information gain is achieved

The intuition behind it is not very hard to grasp:Active Learningemploys humans as domain experts to consistently ‘correct’ or ‘validate’ the outputs from an ML model, with continuous training. This stops when the model achieves acceptable performance. We intuit and rebrand the same, with a few clever innovations that will be detailed in a research pre-print later.

Let’s take a deeper look…

Greedy subset selection with least similar elements

When the number of input labels (classes) is high, the complexity of learning a linear decision boundary between classes increases. As such, the quality of the classifier deteriorates as the number of classes increases. This is especially true when the classifier does not have enoughsamplesto learn from — i.e. each of the training classes has only a few samples.

This is very reflective of a real-world scenario, and the primary motivation behind the creation of R.E.D.

Some ways of improving a classifier’s performance under these constraints:

  • Restrictthe number of classes a classifier needs to classify between
  • Make the decision boundary between classes clearer, i.e., train the classifier onhighly dissimilar classes

Greedy Subset Selection does exactly this — since the scope of the problem is Text Classification, we form embeddings of the training labels, reduce their dimensionality via UMAP, then formSsubsets from them. Each of theSsubsets has elements asntraining labels. We pick training labels greedily, ensuring that every label we pick for the subset is the most dissimilar label w.r.t. the other labels that exist in the subset:

import numpy as np
from sklearn.metrics.pairwise import cosine_similarity


def avg_embedding(candidate_embeddings):
    return np.mean(candidate_embeddings, axis=0)

def get_least_similar_embedding(target_embedding, candidate_embeddings):
    similarities = cosine_similarity(target_embedding, candidate_embeddings)
    least_similar_index = np.argmin(similarities)  # Use argmin to find the index of the minimum
    least_similar_element = candidate_embeddings[least_similar_index]
    return least_similar_element


def get_embedding_class(embedding, embedding_map):
    reverse_embedding_map = {value: key for key, value in embedding_map.items()}
    return reverse_embedding_map.get(embedding)  # Use .get() to handle missing keys gracefully


def select_subsets(embeddings, n):
    visited = {cls: False for cls in embeddings.keys()}
    subsets = []
    current_subset = []

    while any(not visited[cls] for cls in visited):
        for cls, average_embedding in embeddings.items():
            if not current_subset:
                current_subset.append(average_embedding)
                visited[cls] = True
            elif len(current_subset) >= n:
                subsets.append(current_subset.copy())
                current_subset = []
            else:
                subset_average = avg_embedding(current_subset)
                remaining_embeddings = [emb for cls_, emb in embeddings.items() if not visited[cls_]]
                if not remaining_embeddings:
                    break # handle edge case
                
                least_similar = get_least_similar_embedding(target_embedding=subset_average, candidate_embeddings=remaining_embeddings)

                visited_class = get_embedding_class(least_similar, embeddings)

                
                if visited_class is not None:
                  visited[visited_class] = True


                current_subset.append(least_similar)
    
    if current_subset:  # Add any remaining elements in current_subset
        subsets.append(current_subset)
        

    return subsets
ログイン後にコピー

the result of this greedy subset sampling is all the training labels clearly boxed into subsets, where each subset has at most onlynclasses. This inherently makes the job of a classifier easier, compared to the originalSclasses it would have to classify between otherwise!

Semi-supervised classification with noise oversampling

Cascade this after the initial label subset formation — i.e., this classifier is only classifying between a givensubsetof classes.

Picture this: when you have low amounts of training data, you absolutely cannot create a hold-out set that is meaningful for evaluation. Should you do it at all? How do you know if your classifier is working well?

We approached this problem slightly differently — we defined the fundamental job of a semi-supervised classifier to bepre-emptive classification of a sample. This means that regardless of what a sample gets classified as it will be ‘verified’ and ‘corrected’ at a later stage: this classifier only needs to identify what needs to be verified.

As such, we created a design for how it would treat its data:

  • n+1classes, where the last class isnoise
  • noise:data from classes that are NOT in the current classifier’s purview. The noise class is oversampled to be 2x the average size of the data for the classifier’s labels

Oversampling on noise is a faux-safety measure, to ensure that adjacent data that belongs to another class is most likely predicted as noise instead of slipping through for verification.

How do you check if this classifier is working well — in our experiments, we define this as the number of ‘uncertain’ samples in a classifier’s prediction. Using uncertainty sampling and information gain principles, we were effectively able to gauge if a classifier is ‘learning’ or not, which acts as a pointer towards classification performance. This classifier is consistently retrained unless there is an inflection point in the number of uncertain samples predicted, or there is only a delta of information being added iteratively by new samples.

Proxy active learning via an LLM agent

This is the heart of the approach — using an LLM as a proxy for a human validator. The human validator approach we are talking about is Active Labelling

Let’s get an intuitive understanding of Active Labelling:

  • Use an ML model to learn on a sample input dataset, predict on a large set of datapoints
  • For the predictions given on the datapoints, a subject-matter expert (SME) evaluates ‘validity’ of predictions
  • Recursively, new ‘corrected’ samples are added as training data to the ML model
  • The ML model consistently learns/retrains, and makes predictions until the SME is satisfied by the quality of predictions

For Active Labelling to work, there are expectations involved for an SME:

  • when we expect a human expert to ‘validate’ an output sample, the expert understands what the task is
  • a human expert will use judgement to evaluate ‘what else’ definitely belongs to a labelLwhen deciding if a new sample should belong toL

Given these expectations and intuitions, we can ‘mimic’ these using an LLM:

  • give the LLM an ‘understanding’ of what each label means. This can be done by using a larger model tocritically evaluate the relationship between {label: data mapped to label} for all labels. In our experiments, this was done using a32B variant of DeepSeekthat was self-hosted.
R.E.D。:専門家の代表団によるテキスト分類のスケーリング
  • Instead of predicting what is the correct label,leverage the LLM to identify if a prediction is ‘valid’ or ‘invalid’ only(i.e., LLM only has to answer a binary query).
  • Reinforce the idea of what other valid samples for the label look like,i.e., for every pre-emptively predicted label for a sample, dynamically sourcecclosest samples in its training (guaranteed valid) set when prompting for validation.

The result? A cost-effective framework that relies on a fast, cheap classifier to make pre-emptive classifications, and an LLM that verifies these using (meaning of the label + dynamically sourced training samples that are similar to the current classification):

import math

def calculate_uncertainty(clf, sample):
    predicted_probabilities = clf.predict_proba(sample.reshape(1, -1))[0]  # Reshape sample for predict_proba
    uncertainty = -sum(p * math.log(p, 2) for p in predicted_probabilities)
    return uncertainty


def select_informative_samples(clf, data, k):
    informative_samples = []
    uncertainties = [calculate_uncertainty(clf, sample) for sample in data]

    # Sort data by descending order of uncertainty
    sorted_data = sorted(zip(data, uncertainties), key=lambda x: x[1], reverse=True)

    # Get top k samples with highest uncertainty
    for sample, uncertainty in sorted_data[:k]:
        informative_samples.append(sample)

    return informative_samples


def proxy_label(clf, llm_judge, k, testing_data):
    #llm_judge - any LLM with a system prompt tuned for verifying if a sample belongs to a class. Expected output is a bool : True or False. True verifies the original classification, False refutes it
    predicted_classes = clf.predict(testing_data)

    # Select k most informative samples using uncertainty sampling
    informative_samples = select_informative_samples(clf, testing_data, k)

    # List to store correct samples
    voted_data = []

    # Evaluate informative samples with the LLM judge
    for sample in informative_samples:
        sample_index = testing_data.tolist().index(sample.tolist()) # changed from testing_data.index(sample) because of numpy array type issue
        predicted_class = predicted_classes[sample_index]

        # Check if LLM judge agrees with the prediction
        if llm_judge(sample, predicted_class):
            # If correct, add the sample to voted data
            voted_data.append(sample)

    # Return the list of correct samples with proxy labels
    return voted_data
ログイン後にコピー

By feeding the valid samples (voted_data) to our classifier under controlled parameters, we achieve the ‘recursive’ part of our algorithm:

R.E.D。:専門家の代表団によるテキスト分類のスケーリング

By doing this, we were able to achieve close-to-human-expert validation numbers on controlled multi-class datasets. Experimentally, R.E.D. scales up to1,000 classes while maintaining a competent degree of accuracyalmost on par with human experts (90%+ agreement).

I believe this is a significant achievement in applied ML, and has real-world uses for production-grade expectations of cost, speed, scale, and adaptability. The technical report, publishing later this year, highlights relevant code samples as well as experimental setups used to achieve given results.

All images, unless otherwise noted, are by the author

Interested in more details? Reach out to me over Medium or email for a chat!

以上がR.E.D。:専門家の代表団によるテキスト分類のスケーリングの詳細内容です。詳細については、PHP 中国語 Web サイトの他の関連記事を参照してください。

このウェブサイトの声明
この記事の内容はネチズンが自主的に寄稿したものであり、著作権は原著者に帰属します。このサイトは、それに相当する法的責任を負いません。盗作または侵害の疑いのあるコンテンツを見つけた場合は、admin@php.cn までご連絡ください。

ホットAIツール

Undresser.AI Undress

Undresser.AI Undress

リアルなヌード写真を作成する AI 搭載アプリ

AI Clothes Remover

AI Clothes Remover

写真から衣服を削除するオンライン AI ツール。

Undress AI Tool

Undress AI Tool

脱衣画像を無料で

Clothoff.io

Clothoff.io

AI衣類リムーバー

Video Face Swap

Video Face Swap

完全無料の AI 顔交換ツールを使用して、あらゆるビデオの顔を簡単に交換できます。

ホットツール

メモ帳++7.3.1

メモ帳++7.3.1

使いやすく無料のコードエディター

SublimeText3 中国語版

SublimeText3 中国語版

中国語版、とても使いやすい

ゼンドスタジオ 13.0.1

ゼンドスタジオ 13.0.1

強力な PHP 統合開発環境

ドリームウィーバー CS6

ドリームウィーバー CS6

ビジュアル Web 開発ツール

SublimeText3 Mac版

SublimeText3 Mac版

神レベルのコード編集ソフト(SublimeText3)

Meta Llama 3.2を始めましょう - 分析Vidhya Meta Llama 3.2を始めましょう - 分析Vidhya Apr 11, 2025 pm 12:04 PM

メタのラマ3.2:マルチモーダルとモバイルAIの前進 メタは最近、ラマ3.2を発表しました。これは、モバイルデバイス向けに最適化された強力なビジョン機能と軽量テキストモデルを特徴とするAIの大幅な進歩です。 成功に基づいてo

10生成AIコーディング拡張機能とコードのコードを探る必要があります 10生成AIコーディング拡張機能とコードのコードを探る必要があります Apr 13, 2025 am 01:14 AM

ねえ、忍者をコーディング!その日はどのようなコーディング関連のタスクを計画していますか?このブログにさらに飛び込む前に、コーディング関連のすべての問題について考えてほしいです。 終わり? - &#8217を見てみましょう

AVバイト:Meta' s llama 3.2、GoogleのGemini 1.5など AVバイト:Meta' s llama 3.2、GoogleのGemini 1.5など Apr 11, 2025 pm 12:01 PM

今週のAIの風景:進歩、倫理的考慮、規制の議論の旋風。 Openai、Google、Meta、Microsoftのような主要なプレーヤーは、画期的な新しいモデルからLEの重要な変化まで、アップデートの急流を解き放ちました

従業員へのAI戦略の販売:Shopify CEOのマニフェスト 従業員へのAI戦略の販売:Shopify CEOのマニフェスト Apr 10, 2025 am 11:19 AM

Shopify CEOのTobiLütkeの最近のメモは、AIの能力がすべての従業員にとって基本的な期待であると大胆に宣言し、会社内の重大な文化的変化を示しています。 これはつかの間の傾向ではありません。これは、pに統合された新しい運用パラダイムです

ビジョン言語モデル(VLM)の包括的なガイド ビジョン言語モデル(VLM)の包括的なガイド Apr 12, 2025 am 11:58 AM

導入 鮮やかな絵画や彫刻に囲まれたアートギャラリーを歩くことを想像してください。さて、各ピースに質問をして意味のある答えを得ることができたらどうでしょうか?あなたは尋ねるかもしれません、「あなたはどんな話を言っていますか?

GPT-4o vs Openai O1:新しいOpenaiモデルは誇大広告に値しますか? GPT-4o vs Openai O1:新しいOpenaiモデルは誇大広告に値しますか? Apr 13, 2025 am 10:18 AM

導入 Openaiは、待望の「Strawberry」アーキテクチャに基づいて新しいモデルをリリースしました。 O1として知られるこの革新的なモデルは、推論能力を強化し、問題を通じて考えられるようになりました

AIインデックス2025を読む:AIはあなたの友人、敵、または副操縦士ですか? AIインデックス2025を読む:AIはあなたの友人、敵、または副操縦士ですか? Apr 11, 2025 pm 12:13 PM

スタンフォード大学ヒト指向の人工知能研究所によってリリースされた2025年の人工知能インデックスレポートは、進行中の人工知能革命の良い概要を提供します。 4つの単純な概念で解釈しましょう:認知(何が起こっているのかを理解する)、感謝(利益を見る)、受け入れ(顔の課題)、責任(責任を見つける)。 認知:人工知能はどこにでもあり、急速に発展しています 私たちは、人工知能がどれほど速く発展し、広がっているかを強く認識する必要があります。人工知能システムは絶えず改善されており、数学と複雑な思考テストで優れた結果を達成しており、わずか1年前にこれらのテストで惨めに失敗しました。 2023年以来、複雑なコーディングの問題や大学院レベルの科学的問題を解決することを想像してみてください

ラマ3.2を実行する3つの方法-Analytics Vidhya ラマ3.2を実行する3つの方法-Analytics Vidhya Apr 11, 2025 am 11:56 AM

メタのラマ3.2:マルチモーダルAIパワーハウス Metaの最新のマルチモーダルモデルであるLlama 3.2は、AIの大幅な進歩を表しており、言語理解の向上、精度の向上、および優れたテキスト生成機能を誇っています。 その能力t

See all articles