Md. Asif Uddin

    Elementa

    The map

    86 propositions across 74 chapters and 8 books, plus 47 entries in the Apparatus. Every edge below is read from frontmatter. A dependency recorded on one side only cannot exist, because the reverse index is derived rather than written.

    Book I Neural networks

    1. I.1 Computation and RepresentationM2

      feeds I.2 · I.4 · I.5 · I.7 · I.8 · I.9 · II.1 · III.1 · III.3 · III.5 · III.6 · VIII.3

    2. I.2 The PerceptronM3
    3. I.3 Activation FunctionsM3

      feeds I.4 · I.5 · II.6 · III.3

    4. I.4 Loss FunctionsM3
    5. I.5 The Multilayer PerceptronM3

      feeds I.3 · I.6 · I.11

    6. I.6 Backpropagation and Computational GraphsM3

      feeds I.7 · I.8 · I.9 · II.6

    7. I.7 OptimisationM3

      feeds I.8 · II.8

    8. I.8 Normalisation and RegularisationM3

      feeds II.8

    9. I.9 Recurrent Neural NetworksM3

      feeds I.10 · I.11 · II.2 · II.8

    10. I.10 Gated Recurrence: LSTM and GRUM3

      feeds II.2

    11. I.11 Convolutional Neural NetworksM3

      feeds III.2 · III.4 · III.6

    12. I.12 AutoencodersM2

      not yet written

    13. I.13 Variational AutoencodersM3

      not yet written

    14. I.14 Graph Neural NetworksM3

      not yet written

    Book II Foundations

    1. II.1 Tokenisation and EmbeddingsM2

      feeds II.3 · III.6

    2. II.2 From Recurrence to AttentionM2

      feeds II.3

    3. II.3 AttentionM3

      feeds II.4 · II.5 · II.6 · II.7 · III.4

    4. II.4 Multi-Head AttentionM3
    5. II.5 Position and OrderM3

      feeds III.4

    6. II.6 The Transformer BlockM3

      feeds I.7 · III.2 · III.4

    7. II.7 Encoder, Decoder, and Encoder–DecoderM3
    8. II.8 Training at ScaleM3

    Book III Vision

    1. III.1 Images as DataM2

      feeds I.11 · III.4 · III.7

    2. III.2 Convolutional ArchitecturesM2

      feeds III.3

    3. III.3 Representation LearningM1

      feeds III.5 · III.6

    4. III.4 Vision TransformersM3

      feeds III.5

    5. III.5 Self-Supervised VisionM3
    6. III.6 Vision TasksM2
    7. III.7 Medical and Scientific VisionM3

    Book IV Large Language Model (LLM)

    1. IV.1 Language ModellingM3

      not yet written

    2. IV.2 Autoregressive TransformersM3

      not yet written

    3. IV.3 PretrainingM2

      not yet written

    4. IV.4 ScalingM3

      not yet written

    5. IV.5 Instruction Following and AlignmentM3

      not yet written

    6. IV.6 Prompting and In-Context LearningM1

      not yet written

    7. IV.7 Efficient AdaptationM3

      not yet written

    8. IV.8 LLM FailureM2

      not yet written

    9. IV.9 LLM SystemsM2

      not yet written

    Book V Vision-language (VLM)

    1. V.1 Why Join Vision and LanguageM0

      not yet written

    2. V.2 Contrastive Vision-Language LearningM3

      not yet written

    3. V.3 Image–Text AlignmentM2

      not yet written

    4. V.4 Vision-Language GenerationM2

      not yet written

    5. V.5 VLM ArchitecturesM1

      not yet written

    6. V.6 Visual ReasoningM2

      not yet written

    7. V.7 VLM EvaluationM2

      not yet written

    Book VI Causal inference

    1. VI.1 Correlation and CausationM2

      not yet written

    2. VI.2 Causal GraphsM3

      not yet written

    3. VI.3 InterventionsM3

      not yet written

    4. VI.4 IdentificationM3

      not yet written

    5. VI.5 CounterfactualsM3

      not yet written

    6. VI.6 ExperimentsM3

      not yet written

    7. VI.7 Observational StudiesM3

      not yet written

    8. VI.8 Causal Machine LearningM3

      not yet written

    9. VI.9 Causal Reasoning in AIM2

      not yet written

    Book VII Bioinformatics

    1. VII.1 Biological InformationM0

      not yet written

    2. VII.2 Biological DataM2

      not yet written

    3. VII.3 Gene RegulationM2

      not yet written

    4. VII.4 Single-Cell BiologyM3

      not yet written

    5. VII.5 Representation Learning for BiologyM2

      not yet written

    6. VII.6 PerturbationM3

      not yet written

    7. VII.7 Biological Sequences as GraphsM3

      not yet written

    8. VII.8 Phylogenetics and Evolutionary GraphsM3

      not yet written

    9. VII.9 Biological NetworksM2

      not yet written

    10. VII.10 Molecular Graphs and Drug DiscoveryM2

      not yet written

    Book VIII Practice

    1. VIII.1 Reading PapersM0

      not yet written

    2. VIII.2 Problem FormulationM1

      not yet written

    3. VIII.3 Dataset PracticeM2

      feeds III.7

    4. VIII.4 Experimental DesignM2

      not yet written

    5. VIII.5 EvaluationM3

      not yet written

    6. VIII.6 Training SystemsM3

      not yet written

    7. VIII.7 Debugging ModelsM1

      not yet written

    8. VIII.8 ReproducibilityM0

      not yet written

    9. VIII.9 Research CommunicationM0

      not yet written

    10. VIII.10 Research ProjectsM1

      not yet written

    Unattached

    A proposition that depends on nothing and that nothing depends on is either the first idea in a chain or a page that has not been wired in. The graph guard reports these; it does not fail on them, because some of them are genuinely first.

    None. Every proposition sits on an edge.