The Challenges of Machine Learning: A Critical Review

Abstract

The concept of learning has multiple interpretations, ranging from acquiring knowledge or skills to constructing meaning and social development. Machine Learning (ML) is considered a branch of Artificial Intelligence (AI) and develops algorithms that can learn from data and generalize their judgment to new observations by exploiting primarily statistical methods. The new millennium has seen the proliferation of Artificial Neural Networks (ANNs), a formalism able to reach extraordinary achievements in complex problems such as computer vision and natural language recognition. In particular, designers claim that this formalism has a strong resemblance to the way the biological neurons operate. This work argues that although ML has a mathematical/statistical foundation, it cannot be strictly regarded as a science, at least from a methodological perspective. The main reason is that ML algorithms have notable prediction power although they cannot necessarily provide a causal explanation about the achieved predictions. For example, an ANN could be trained on a large dataset of consumer financial information to predict creditworthiness. The model takes into account various factors like income, credit history, debt, spending patterns, and more. It then outputs a credit score or a decision on credit approval. However, the complex and multi-layered nature of the neural network makes it almost impossible to understand which specific factors or combinations of factors the model is using to arrive at its decision. This lack of transparency can be problematic, especially if the model denies credit and the applicant wants to know the specific reasons for the denial. The model’s “black box” nature means it cannot provide a clear explanation or breakdown of how it weighed the various factors in its decision-making process. Secondly, this work rejects the belief that a machine can simply learn from data, either in supervised or unsupervised mode, just by applying statistical methods. The process of learning is much more complex, as it requires the full comprehension of a learned ability or skill. In this sense, further ML advancements, such as reinforcement learning and imitation learning denote encouraging similarities to similar cognitive skills used in human learning.

Machine Learning, Mechatronics, and Stretch Forming: A History of Innovation in Manufacturing Engineering

“…Recent developments in machine learning, including reinforcement learning and imitation learning, suggest a growing resemblance to the cognitive skills used in human learning. [184]. The potential of artificial intelligence (AI) and machine learning (ML) for solving practical problems is immense.…”

Bibliometric Analysis of Intelligent Systems for Early Anomaly Detection in Oil and Gas Contracts: Exploring Recent Progress and Challenges

“…(2) the requirement for a deep contextual understanding to discern anomalies where contextual nuances might otherwise mask them accurately; (3) the scarcity of labeled datasets which hampers supervised learning in natural language processing; (4) the dynamic nature of language and the concept of 'anomalies' which complicate consistent identification; and (5) the complication of meaningful anomaly detection due to noise and data quality issues, the necessity of models that can adapt to intentional text manipulation by malicious actors seeking to bypass anomaly detection, and the computational strain posed by processing large volumes of text in real time which poses challenges in maintaining the balance between speed and accuracy under resource constraints [60][61][62]. Further investigations will be performed on the literature exploring the impact of different approaches involving natural language processing (NLP) on predictive tasks in public procurement calls, particularly examining the performance of language-agnostic models like LaBSE (Language-agnostic BERT Sentence Embedding) and LASER (Language-Agnostic SEntence Representations) to enhance prediction accuracy.…”

“…Figure 11 provides a concise overview of key obstacles faced in anomaly detection within textual content in machine learning. These encompass the following drawn conclusions: (1) the need for models sensitive to linguistic and cultural variations due to different norms across languages and cultures; (2) the requirement for a deep contextual understanding to discern anomalies where contextual nuances might otherwise mask them accurately; (3) the scarcity of labeled datasets which hampers supervised learning in natural language processing; (4) the dynamic nature of language and the concept of anomalies' which complicate consistent identification; and (5) the complication of meaningful anomaly detection due to noise and data quality issues, the necessity of models that can adapt to intentional text manipulation by malicious actors seeking to bypass anomaly detection, and the computational strain posed by processing large volumes of text in real time which poses challenges in maintaining the balance between speed and accuracy under resource constraints [60][61][62].…”

Objective Model Selection in Physics: Exploring the Finite Information Quantity Approach

“…The abundance of parameters in complex models can make them susceptible to overfitting, leading to unreliable predictions. Therefore, careful evaluation and techniques like regularization are crucial to ensure model generalizability [24].…”