TestSTMemory vs Learn Deep Learning

* The app allows you to assess the level of short-term visual memory. The algorithm of the program was as follows: the test participant had to complete five stages, each consisting of ten attempts. At the first stage, during the first five attempts, the participant had to react to a single monochrome signal, memorize its location, and click on the corresponding circle. In the next five attempts, they had to respond to a colored signal. At each subsequent stage, the number of simultaneously appearing signals increased by one. By the fifth stage, the participant needed to memorize the locations of five signals and click on the corresponding circles. During the test, participants had to respond quickly and accurately to visual stimuli. The app's display showed the number of taps already made and how many remained (for stages 3, 4, and 5). To objectively assess short-term visual memory, the percentage of errors made during the test was calculated. A mistake was defined as clicking on a circle that did not correspond to the correct one. At the end of the test, the program prompted the user to enter information about the test participant and displayed indicators characterizing the measured quality." * Model characteristics could be generated based on the measurement results. To do this, the user needed to navigate to the 'Data' section, click 'Choose', select the desired measurements, and then click 'Model'. On the next screen, the main characteristics of the selected measurements were displayed. If the characteristics were satisfactory, the user could click 'Create model'. Created models were accessible in the settings under 'Models based on measurements'. To compare measurement results with a model, the user had to go to the 'Data' section, select a measurement, click 'Summary', and on the next screen, select 'Assessment'. As an example, the model characteristics of martial artists with high sports qualifications ('The model of qualified martial artists') were set by default." * Additionally, in the settings, users could save a backup of their measurements in JSON format ('Save backup'), import measurements from another device ('Add backup'), replace existing measurements with new ones ('Replace backup'), or delete all data ('Delete all data'). * The “Short-Term Visual Memory” app includes a machine learning model (Core ML) that automatically determines the level of short-term visual memory efficiency based on the results of a 5-stage test. The model analyses accuracy and attempt duration at stages 3, 4, and 5, because these stages involve higher cognitive load and better differentiate athletes by their ability to retain and reproduce visual stimuli. The model was developed using test results from combat sports athletes (n = 338) obtained under the following standardised task settings: number of cells — 77; visual stimulus duration — 300 ms. The result is presented as a short-term visual memory efficiency level (High / Medium / Low) and an integral STM Efficiency Index. This allows the app to be used for regular monitoring of athletes’ cognitive and psychophysiological functions, tracking individual changes over time, and selecting training tasks according to the athlete’s current condition. All calculations are performed locally on the device, without transmitting personal data to a server. Test results should be interpreted considering the athlete’s level of preparedness, testing conditions, and repeated-measurement dynamics. * The app is not a medical device and is not intended to provide medical diagnoses.

Master Deep Learning, Neural Networks, and Artificial Intelligence with the most comprehensive learning app. From Python fundamentals to training Large Language Models (LLMs), building Generative AI apps, and deploying models to the cloud — this is your complete path to becoming an AI Engineer. COMPLETE CURRICULUM - Zero to AI Expert.Start from scratch and become job-ready with our structured learning path: Deep Learning Fundamentals: • Introduction to Deep Learning & Neural Networks • How Deep Learning differs from traditional Machine Learning • Artificial Neural Networks (ANN) architecture & training • Loss Functions, Optimizers (Adam, SGD, RMSprop) • Regularization: Dropout, Batch Norm, L1/L2 Core Architectures: • Convolutional Neural Networks (CNNs) for Computer Vision • Recurrent Neural Networks (RNNs), LSTM & GRU for sequences • Transformers: Attention mechanism • Generative Models: GANs, VAEs, Diffusion Models • Autoencoders & Self-Supervised Learning Libraries & Frameworks: • PyTorch: Tensors, Autograd, nn.Module, training loops • TensorFlow & Keras: Sequential, Functional API, model deployment • JAX: High-performance ML research framework • NumPy, Pandas, Scikit-learn, SciPy foundations Computer Vision: • Image Classification with CNNs • Object Detection: YOLO, R-CNN, SSD • Image Segmentation: U-Net, Mask R-CNN • Face Recognition & Facial Landmarks MLOps & Engineering: • ML lifecycle management & experiment tracking • MLflow, Weights & Biases (W&B) for logging • Kubeflow pipelines & Apache Airflow orchestration • CI/CD for Machine Learning • Model versioning, registries & A/B testing • Monitoring, drift detection & model retraining Production & Deployment: • FastAPI & Flask for serving ML models • gRPC for high-performance model inference • Docker containerization for ML apps • Load balancing, caching & request queuing • PostgreSQL, MongoDB & Redis integration • Vector databases: Pinecone, Weaviate, Milvus for RAG Cloud Infrastructure: • AWS SageMaker, EC2, Lambda for ML workloads • Azure Machine Learning platform • Kubernetes (K8s) & Helm for ML orchestration Hardware & Compute • NVIDIA GPUs, CUDA cores & Tensor cores • Neural Processing Units (NPU) & Apple Neural Engine Edge & Mobile AI: • TensorFlow Lite for mobile deployment • Core ML for iOS apps • ONNX for cross-platform model export AI Security & Ethics: • AI safety & alignment fundamentals • Adversarial attacks & robustness • Data poisoning & model stealing defenses • Fairness, bias detection & mitigation Specialized Domains: • Medical AI: Diagnosis, radiology, drug discovery • Robotics: Perception, control, reinforcement learning • Financial AI: Trading, fraud detection, risk modeling • Bioinformatics: Protein folding, genomics • Cybersecurity: Threat detection, anomaly detection Research & Expert Topics: • Foundation models, scaling laws & emergent abilities • Mixture of Experts (MoE) & Sparse Transformers • World models, neuro-symbolic AI • Meta-learning, continual learning, curriculum learning AI TUTOR - Your 24/7 Learning Assistant: • Ask any Deep Learning or AI question • Get explanations of neural network architectures • Debug model training issues with AI help GAMIFIED LEARNING - Stay Motivated: • Daily learning streaks with fire animations • XP points & level progression • Study reminders with push notifications POWERFUL ORGANIZATION TOOLS: • Bookmarks: Save lessons for quick access • Notes: Write personal notes on any lesson • Search: Find anything instantly across all lessons • Dark mode for comfortable night learning LEARN OFFLINE - Anytime, Anywhere: • All content are offline access • Study on your commute without internet PERFECT FOR: • Aspiring AI & Machine Learning Engineers • Data Scientists expanding into Deep Learning • Software developers transitioning to AI • Students studying computer science or AI • Career changers entering the AI field Start your Deep Learning Journey today !