Here are some notes on papers I have read.

Here are some notes for COMP90084.

Here are some candidate research topics for second-year in UniMelb.

Resource on DL/AI (Maybe watch/listen the seminars/podcasts first, then the online courses and books)

• Books & Online Course Materials:

  • Understanding Deep Learning

  • Artificial Intelligence: A Modern Approach

  • Reinforcement Learning: An Introduction

  • What Is ChatGPT Doing…and Why Does It Work?

  • An Introduction to Universal Artificial Intelligence

  • Stanford CS25: Transformers United

  • Stanford CS224n: Natural Language Processing with Deep Learning, YouTube Playlist

  • Neural Networks: Zero to Hero by Andrej Karpathy (former cofounder of OpenAI)

  • Hugging Face Deep Reinforcement Learning Course

  • Hugging Face Diffusion Models Course

  • Christopher Olah’s Blog

  • Transformer circuits thread

• Lilian Weng’s blog, such as

  • What are Diffusion Models?

  • Prompt Engineering

  • LLM Powered Autonomous Agents

• Coursera/deeplearning.ai courses and specialisations:

  • How Diffusion Models Work

  • Machine Learning Specialization

  • Deep Learning Specialization

  • Natural Language Processing Specialization

• Videos/Podcasts/Seminars:

  • Yannic Kilcher’s channel. He makes paper-reading videos

  • Machine Learning Street Talk. These people interview researchers in AI. Interesting ones:

  • Neural and Non-Neural AI, Reasoning, Transformers, and LSTMs

  • Prof. Bob Coecke – Quantum Natural Language Processing

  • The AI Bubble: Will It Burst, and What Comes After?

  • Podcast from DeepMind

  • Compression for AGI – Jack Rae

  • Yann LeCun: Meta AI, Open Source, Limits of LLMs, AGI & the Future of AI | Lex Fridman Podcast

  • Yann LeCun | Objective-Driven AI: Towards AI systems that can learn, remember, reason, and plan

  • Geoffrey Hinton | On working with Ilya, choosing problem, and the power of intuition. Not really on specific research topics, but talks about Hinton’s experience with AI in general

  • Geoffrey Hinton – Two Paths to Intelligence

  • AlphaGo – The Movie | Full award-winning documentary

• Papers/Reviews/Preprints/Articles/News Release:

  • The Bitter Lesson

  • Physics of Language Models

  • The Alberta Plan for AI Research

  • A Path Towards Autonomous Machine Intelligence

  • A Primer on the Inner Workings of Transformer-based Language Models

  • Understanding Diffusion Models: A Unified Perspective

  • Step-by-Step Diffusion: An Elementary Tutorial

  • The statistical thermodynamics of generative diffusion models: Phase transitions, symmetry breaking and critical instability

  • Deep Generative Modelling: A Comparative Review of VAEs, GANs, Normalizing Flows, Energy-Based and Autoregressive Models

  • Designing Ecosystems of Intelligence from First Principles

  • Compositional Generative Modelling: A Single Model is Not All You Need

  • The Shift from Models to Compound AI Systems

  • The Landscape of Emerging AI Agent Architectures for Reasoning, Planning, and Tool Calling: A Survey

• …

What is Observables？

(Page 125 of the textbook)

• Observables are represented by hermitian operators. The result of an observation is always an eigenvalue of the hermitian.
• The expression represents the expected value of observing on .
• Observables in general do not commute. This means that the order of observation matters. Moreover, if the commutator of two observables is not zero, there is an intrinsic limit to our capability of measuring their values simultaneously.

Prerequisites 1. Introduction to Genomics 基因组学

• eukaryotes, prokaryotes, viruses
  • 真核生物、原核生物、病毒
  • water 70%, protein 15%, RNA 6%, DNA 1%
• genome 基因组
  • bacterium, human, plant, virus
  • 细菌、人类、植物、病毒
  • chromosome 染色体
  • plasmid 质粒
  • mitochondria 线粒体
  • chloroplast 叶绿体
• ploidy 倍性
  • 2 copies of the same chromosome, one from each parent
  • haploid 单倍体
  • diploid 二倍体
  • polyploid, triploid, tetraploid
• molecular biology 分子生物学
  • DNA, RNA, protein
  • nucleotide 核苷酸, amino acid 氨基酸
  • gene, genome, transcriptome 转录组, proteome 蛋白质组
• DNA: A-T, C-G
  • double helix 双螺旋
  • sugar-phosphate backbone 糖磷酸骨架
  • 5' to 3' direction
  • chromosome, gene, exon 外显子, intron 内含子, promoter 启动子, enhancer 增强子, terminator 终止子
• protein
  • encoded by gene
  • polymer of amino acid monomers 氨基酸单体聚合物
  • cellular functions 细胞功能
    • catalysis 催化
    • structure 结构
    • signalling 信号
• amino acid
  • 20 types
  • 3-letter and 1-letter code
  • hydrophobic 疏水, hydrophilic 亲水, charged 带电
• central dogma
  • DNA -> RNA -> protein
  • transcription 转录, translation 翻译
  • U (uracil) in RNA instead of T (thymine) in DNA
  • ribosome 核糖体, tRNA, mRNA
• reserve transcription 逆转录
  • mRNA -> DNA
  • reverse transcriptase 逆转录酶
  • retrovirus 逆转录病毒
• gene
  • unit of heredity 遗传单位
  • 2% of human genome codes for genes
  • promoter, 5' UTR, coding sequence, 3' UTR, terminator
  • coding sequence: exon, intron
• variation
  • sequence variation
    • single nucleotide variation (SNV) 单核苷酸变异
    • single nucleotide polymorphism (SNP) 单核苷酸多态性
    • indels: insertion, deletion
  • structural variation
    • large-scale (>50bp)
    • involving entire sections of DNA
    • gain or loss of fucntion (GoF, LoF)
  • conservation
• sequencing technology
  • DNA/RNA extraction -> library preparation -> sequencing -> export fastq files -> QC/analysis
  • Illumina, Nanopore
• FASTQ files, reads
  • 4 lines per read
    • identifier - unique name
    • sequence - AGTC
    • spacer - +
    • quality - confidence score for each DNA base
• sequencing strategies
  • whole genome sequencing (WGS)
    • all DNA in cells
  • whole exome sequencing (WES)
    • all exons
    • cheap, capture importants
  • RNA sequencing (RNA-seq)
    • all RNA in cells
  • scRNA-seq
    • all RNA but idividual cells
  • Amplicon
    • use PCR target specific region
  • Pair-end
    • sequence both ends of DNA fragment

What is this course about?

This course is an introduction to natural language processing for engineering students. The course covers the following topics:

1. Preprocessing

• sentence segmentation
• tokenization, subword tokenization
• word normalization
  • inflectional vs derivational morphology
  • lemmatization vs stemming
• stopword removal

What is this course about?

This course is an introduction to research methods for computer science students.

What is research?

In a research domain, come up with a research question, and then find the answer to the question. A good statement of the research question is the half of the answer.

Does any resource help with this course?

• Mastering Reinforcement Learning
• 蘑菇书EasyRL
• JoyRL Book

Does any resource help with this course?

1. Algorithm Design by Kleinberg and Tardos
2. Algorithms by Jeff Erickson
3. Linear Programming
4. Design of Approximation Algorithms