#cross-entropy

[fantasblog]

Scissorx:Their parent got divorced...

C-E: yes

Scissorx: are you sure?

C-E: yeah,Nerd Lil bro

Credit:

Scissor x (Errorcrink shipchild and little brother of cross-entropy)belong to me

Cross-entropy(errorcrink shipchild and older brother of Scissor x) belongs to @litter-bag

× DoNt REPOST mY aRt ×

(Scissor x head takes 1 hours)

That all -w-

[ixnai]

The Mirage of AI: Blandishment and Hallucination in Autoregressive Models

TL;DR: Autoregressive models often falter in long sequences, where blandishment and hallucination arise from sampling failures, challenging the reliability of perplexity and cross-entropy as accuracy metrics.

AI models can lie. Not intentionally, of course, but through a process of blandishment and hallucination that emerges from autoregressive sampling failures. These failures, particularly in long sequences, reveal the limitations of current AI systems and the metrics we use to evaluate them.

Autoregressive models, like those used in many AI applications, predict the next token in a sequence based on previous tokens. This process, however, is fraught with potential errors, especially when using temperature-scaled softmax sampling. As the model generates longer sequences, small errors can compound, leading to significant deviations from factual accuracy. This is where blandishment—overly flattering or misleading output—and hallucination—entirely fabricated content—come into play.

Perplexity and Cross-Entropy: These are standard metrics for evaluating language models, but they fall short in assessing factual accuracy. Perplexity measures how well a model predicts a sample, while cross-entropy evaluates the difference between predicted and actual distributions. Neither metric accounts for the truthfulness of the content, allowing models to produce plausible yet incorrect information.

Sampling Techniques: Beam search, nucleus sampling, and top-k sampling each have their own failure modes. Beam search can lead to repetitive and uncreative outputs, while nucleus and top-k sampling may introduce randomness that exacerbates hallucination. Each method struggles to balance creativity with accuracy.

Information Theory and Log-Likelihood: Maximizing log-likelihood is a common training objective, yet it doesn’t ensure semantic coherence or truthfulness. Information theory suggests that while a model may be statistically optimal, it can still produce semantically incoherent or false outputs.

Attention Entropy: This metric can help detect when models are ‘guessing.’ High entropy in attention weights indicates uncertainty, often correlating with less reliable outputs. Monitoring attention entropy could provide a warning system for potential inaccuracies.

In the wake of recent AI funding bubbles and overpromised capabilities, it’s crucial to scrutinize these models more closely. As we continue to integrate AI into critical areas, from healthcare to finance, ensuring the semantic coherence and truthfulness of AI outputs is paramount. How can we refine our models and metrics to better align with these goals?

[doyouknowkrwp]

AI 모델 성능 평가 지표 완전 정복: F1부터 Policy Gradient까지, 숫자 하나가 모델의 운명을 결정한다

모델을 학습시켰는데 "좋다"는 게 뭔가요?정확도 99%가 항상 최고일까요? 암 진단 AI가 정확도 99%라며 자랑하는데, 실제 암 환자 100명 중 90명을 놓친다면? 같은 모델인데 평가 지표만 바꿨더니 최고에서 최악으로 추락하는 충격적인 현실! MSE, RMSE, MAE부터 AUROC, F1-Score, mAP, BLEU, ROUGE, Perplexity, IoU, FID, CLIPScore, BERTScore까지, 데이터 타입별·Task별·응용 분야별로 어떤 지표를 써야 하는지 완벽 정리했습니다. #Accuracy #AI평가지표 #AUROC #CrossEntropy #DiceLoss #F1Score #PolicyGradient #REINFORCE #sigmoidF1 #SurrogateLoss #강화학습 #머신러닝평가 #미분가능 #손실함수 Read the full article

[doyouknowkrwp]

AI 학습의 숨은 심판관, 손실함수 완전 정복: MSE·Cross-Entropy부터 Focal Loss·Huber Loss까지 (회귀/분류별 선택 가이드)

"모델 성능이 안 나올 때 첫 번째로 의심해야 할 것!" MSE vs MAE vs Huber Loss, Binary vs Categorical Cross-Entropy, Focal Loss, Dice Loss까지. 회귀/분류 문제별 최적 손실함수 선택법, PyTorch/TensorFlow 구현 코드, 실전 문제 해결 전략까지 완벽 가이드. #CrossEntropy #FocalLoss #LossFunction #MSE #PyTorch #TensorFlow #딥러닝 #머신러닝 #손실함수 Read the full article

[biomedres]

Class Based Variable Importance for Medical Decision Making

Class Based Variable Importance for Medical Decision Making by Sanjay Warrier in Biomedical Journal of Scientific & Technical Research (BJSTR) https://biomedres.us/fulltexts/BJSTR.MS.ID.000431.php

For more Biomedical open access journals please click on

[haohanwang]

My first trained Multinomial logistic regression Model o tensorflow