Today, both science and industry rely heavily on machine learning models, predominantly artificial neural networks, that become increasingly complex and demand more computing resources to be trained. In this paper, we will look holistically at the efficiency of machine learning models and draw the inspirations to address their main challenges from the green sustainable economy principles. Instead of constraining some computations or memory used by the models, we will focus on reusing what is available to them: computations done in the previous processing steps, partial information accessible at run-time, or knowledge gained by the model during previous training sessions in continually learned models. This new research path of zero-waste machine learning can lead to several research questions related to efficiency of contemporary neural networks - how machine learning models can learn better with less data? How they select relevant data samples out of many? Finally, how can they build on top of already trained models to reduce the need for more training samples? Here, we explore all the above questions and attempt to answer them.

Increasing societal awareness of mental health challenges has significantly reduced stigma surrounding psychological disorders, encouraging greater numbers of individuals to seek professional support, which has placed unprecedented pressure on mental health services, with institutions ranging from educational establishments to emergency services implementing systematic screening protocols to identify individuals requiring intervention. However, the growing demand for rapid, accurate diagnosis continues to strain limited professional resources. Our study introduces an innovative machine learning framework for mental disorder detection using electroencephalography (EEG) signals processed through Welch’s power spectral density estimation. Unlike conventional Fast Fourier Transform (FFT) approaches, our method generates refined two-dimensional spectrograms capturing brain wave amplitudes (in dB) alongside precise peak frequency identification. This computationally efficient periodogram variant enables robust feature extraction suitable for real-time diagnostic applications while reducing model training overhead. Preliminary analysis demonstrates the Welch Transform’s superior signal characterization compared to standard FFT periodograms, revealing distinct neurophysiological patterns associated with various mental health conditions. The approach maintains high computational efficiency, supporting potential deployment in clinical screening environments.

In this work, we introduce a novel approach for optimizing LLM training by adjusting learning rates across weights of different components in Transformer models. Traditional methods often apply a uniform learning rate across all network layers, potentially overlooking the unique dynamics of each part. Remarkably, our introduced relative learning rates, RLRS, method accelerates the training process by up to $23\%$, particularly in complex models such as Mixture of Experts (MoE). Hyperparameters of RLRS can be efficiently tuned on smaller models and then effectively reused on models up to $27\times$ larger. This simple and effective method results in a substantial reduction in training time and computational resources, offering a practical and scalable solution for optimizing large-scale neural networks.