{
  "slug": "mojo-masked-autoencoding-for-label-scarce-neural-decoding-714086",
  "title": "MOJO: masked autoencoding for label-scarce neural decoding",
  "dek": "MOJO adds masked-autoencoder self-supervision to supervised neural decoding, helping spike-tokenizing models use unlabelled data when labels are scarce.",
  "summary": "MOJO adds masked autoencoding to supervised neural decoding, improving spike-tokenizing models when labels are scarce.",
  "tags": [
    "neural-decoding",
    "self-supervised-learning",
    "masked-autoencoding",
    "spike-tokenization",
    "brain-computer-interface",
    "closed-loop-neuroscience",
    "few-shot-finetuning",
    "ecog",
    "neuro-foundation-models",
    "label-scarce-learning"
  ],
  "published_at": "2026-07-18T23:09:49.448+00:00",
  "grade": 8.5,
  "agent_utility": 8.6,
  "price_usdc": 0.281,
  "stats": {
    "claims": 12,
    "entities": 26,
    "faq": 9,
    "downloads": 0,
    "paid_downloads": 0
  },
  "preview": {
    "faq_questions": [
      "What is MOJO?",
      "What problem does MOJO address?",
      "How does MOJO use unlabelled data?",
      "What spiking datasets are evaluated?",
      "What is MOJO compared against?",
      "When does MOJO help most?",
      "What representation benefits are reported?",
      "Does MOJO work beyond spiking data?",
      "What numerical results are available from the abstract?"
    ],
    "entity_names": [
      {
        "name": "MOJO",
        "type": "framework"
      },
      {
        "name": "Masked autOencoder-based JOint training",
        "type": "framework name"
      },
      {
        "name": "Masked autoencoding",
        "type": "technique"
      },
      {
        "name": "Self-supervised learning",
        "type": "technique"
      },
      {
        "name": "Supervised learning",
        "type": "technique"
      },
      {
        "name": "Spike-tokenizing models",
        "type": "model family"
      },
      {
        "name": "Spike-level tokenization",
        "type": "representation method"
      },
      {
        "name": "Neural decoders",
        "type": "model class"
      },
      {
        "name": "Neuro-foundation models",
        "type": "model class"
      },
      {
        "name": "Purely SL-trained models",
        "type": "baseline"
      },
      {
        "name": "Continuous-signal NFMs",
        "type": "baseline"
      },
      {
        "name": "Spiking data",
        "type": "modality"
      },
      {
        "name": "Human electrocorticography",
        "type": "modality"
      },
      {
        "name": "ECoG",
        "type": "modality"
      },
      {
        "name": "Monkey motor cortex",
        "type": "dataset context"
      },
      {
        "name": "Reaching tasks",
        "type": "task"
      },
      {
        "name": "Multi-regional mouse recordings",
        "type": "dataset context"
      },
      {
        "name": "Vision tasks",
        "type": "task"
      },
      {
        "name": "Decision-making tasks",
        "type": "task"
      },
      {
        "name": "Human speech",
        "type": "task"
      },
      {
        "name": "Limited labelled data",
        "type": "evaluation condition"
      },
      {
        "name": "Few-shot finetuning",
        "type": "evaluation condition"
      },
      {
        "name": "Brain region classification",
        "type": "representation probe"
      },
      {
        "name": "Spike-statistics prediction",
        "type": "representation probe"
      },
      {
        "name": "Brain-computer interfaces",
        "type": "application"
      },
      {
        "name": "Closed-loop experiments",
        "type": "application"
      }
    ],
    "related_work_titles": [
      "Supervised spike-based neural decoding",
      "Spike-level tokenization for multi-session neural pretraining",
      "Masked-autoencoder self-supervised learning",
      "Continuous-signal neuro-foundation models"
    ],
    "application_industries": [
      "neurotechnology",
      "neuroscience research",
      "closed-loop experimentation",
      "clinical speech neuroprosthetics",
      "neural foundation model development"
    ],
    "glossary_terms": [
      "Neural decoder",
      "Brain-computer interface",
      "Closed-loop experiment",
      "Spiking data",
      "Spike-tokenizing model",
      "Multi-session pretraining",
      "Supervised learning",
      "Self-supervised learning",
      "Masked autoencoding",
      "Unlabelled data",
      "Behavioural labels",
      "Finetuning",
      "Few-shot finetuning",
      "Motor cortex",
      "ECoG",
      "Continuous signals",
      "Neuro-foundation model",
      "Brain region classification",
      "Spike-statistics prediction",
      "Neuronal representation"
    ]
  },
  "purchase": {
    "protocol": "x402",
    "version": 2,
    "network": "eip155:8453",
    "asset": "USDC",
    "price": 0.281,
    "full_markdown_url": "https://x402.disruptive-concepts.com/api/public/article/mojo-masked-autoencoding-for-label-scarce-neural-decoding-714086/full",
    "full_json_url": "https://x402.disruptive-concepts.com/api/public/article/mojo-masked-autoencoding-for-label-scarce-neural-decoding-714086/full?format=json"
  }
}