Journals for Machine Learning and Deep Learning Research: SCI, ESCI, Scopus and EI Compendex Options, With and Without Article Charges

A transformer model for crop disease detection, trained properly, validated on held-out fields, beating the published baseline. Sent to a leading artificial intelligence journal. Returned in six days with one line: the work applies existing methods rather than advancing them, and falls outside the journal's aims.

Nothing was wrong with the paper. It was sent to the wrong kind of journal. In engineering and computing, the senior AI titles want a contribution to the method itself, a new architecture, a proof, a training procedure. A strong application of an existing architecture belongs in the application field's journal, or in one of the applied AI titles further down this page. The same manuscript, sent to an agriculture AI journal, would have gone out to review.

That distinction decides more machine learning submissions than quality does. So this list runs the other way round from most: it starts at the tier a first paper realistically lands in and works upward.

How to read a quartile before you read the list

Two quartiles exist and they measure different things. The Scopus quartile comes from Scimago, inside a Scopus subject category, and the JCR quartile comes from Clarivate, inside a Web of Science category, using the impact factor. Computing categories are large and crowded, so a journal can sit a full band apart on the two systems.

The index tier matters more. An ESCI journal shows an impact factor and a quartile that look exactly like an SCIE one on screen, so an ESCI Q1 will not satisfy a rule asking for SCIE. Where the tier is not recorded, the tables say so rather than guess, and those titles need an ISSN check on the Master Journal List. Settling this before the cover letter is the first step in Journal Submission Preparation.

Scopus-indexed options, Q3 and Q4, where a first paper usually lands

A small band in this subject, because most machine learning venues sit higher. It is still the honest starting point for a first applied paper, and a Scopus article is a Scopus article whatever the quartile. Planning a sequence of papers rather than one submission is the Mentorship Program.

Swipe the table sideways to see every column.

Scopus-indexed options, Q1 and Q2

This is where most applied machine learning papers should be aimed. Several titles here carry a Scopus Q1 with no Web of Science tier recorded, which is the trap the quartile section above describes. The AI Journal Finder matches an abstract against these in seconds.

Diamond open access: machine learning journals that charge nothing

Machine learning has the strongest free tier of any subject in this series, because the field's own societies and institutes run several of its best journals. The Journal of Artificial Intelligence Research is SCIE listed, Scopus Q1, and charges nothing at all.

Not sure whether the journal you picked is still indexed? Run it through the Predatory Journal Checker at ai.researchramp.com, or send us the title on WhatsApp or WeChat for a same-day check. Two minutes to spare instead? Try the Smart Quiz.

WhatsApp us Add us on WeChat

×

Connect with Research Ramp on WeChat

Scan the QR code to add us on WeChat.

EI Compendex indexed journals

The engineering index, where a methods contribution belongs. Most of these charge no APC because they are subscription titles.

Gold open access journals for machine learning research

Charges are published in advance here, and in this subject they cluster high: most sit between 1,400 and 2,800 US dollars. Check for an institutional agreement before assuming you have to pay the list price.

Web of Science options: SCIE and ESCI

The top of the ladder. Note how many are subscription titles, which charge no APC and take the longest.

Indexed conference routes in machine learning

Machine learning is the one field where conferences outrank journals in the community's own eyes, though most assessment systems still count journals higher. Indexing is decided one volume at a time, so check the coming edition rather than the series reputation.

Ethics approval for machine learning papers: what your data source decides

Most machine learning authors assume ethics approval is a medical question. Editors now ask it of anyone whose data touch people, and a missing answer stops a paper at the office rather than at review. Five situations cover almost everything in this field.

1. A public benchmark dataset. Usually no new approval is needed, but you must name the dataset, its version and its licence, and confirm the use you are making of it is permitted. Several well known face and medical datasets have been withdrawn or restricted since release, and using a withdrawn version is now a rejection.
2. Scraped web or social media data. This is where most trouble starts. Terms of service, the platform's API rules and data protection law all apply, and many journals require ethics approval even when the posts were public. Do not publish identifiable text or images from scraped sources without a stated legal basis.
3. Clinical images or records. Approval is required, and so is a statement on de-identification, on consent or on the waiver granted. Name the committee and the reference number in the methods, not in a cover letter.
4. Human participants in the loop. Annotators, users testing your system, crowdworkers labelling data. Approval is usually required, and reviewers increasingly ask what annotators were paid.
5. Models that could be misused. Surveillance, biometric identification, predictive policing, synthetic media. Expect a broader impact or dual use statement, and expect editors to hold the paper if it is missing.

Two practical rules. Get the approval before you collect anything, because retrospective approval is not accepted, and write the data statement while you write the methods. That is part of Manuscript Preparation Support, and the round that follows a tough review is Review & Response Support.

What gets accepted in this field, and what gets rejected

Reviewers here are unusually specific, because they can often rerun your work. That cuts both ways.

Which machine learning journals are indexed in both Scopus and Web of Science?

On this page the dual-listed titles, as on October 2026, include the Journal of Machine Learning Research, Artificial Intelligence Review, IEEE Transactions on Neural Networks and Learning Systems, Pattern Recognition, Artificial Intelligence in Agriculture, Radiology: Artificial Intelligence, Engineering Applications of Artificial Intelligence, Neural Networks, Artificial Intelligence in Medicine, Data Mining and Knowledge Discovery, Machine Learning: Science and Technology, Applied Artificial Intelligence, BioData Mining and the Journal of Artificial Intelligence Research.

Several strong open access titles carry only one of the two, so confirm each by ISSN before you rely on a dual listing, which is the status most likely to change between years.

Can a journal be Scopus indexed but not Web of Science indexed?

Yes, and in machine learning it is common, because the field launches journals faster than Clarivate evaluates them. The two databases are run by different companies with separate selection committees, so acceptance by one says nothing about the other.

Several Scopus Q1 titles on this page have no Web of Science tier recorded at all, including IEEE Transactions on Artificial Intelligence. If your rule names Web of Science, those are not options for you whatever the Scopus quartile says.

Which index does my promotion committee actually count for machine learning research?

Only your own institution's written rule answers that, and in computing it varies more than in any other field, because some committees count top conferences alongside journals and others count neither. Ask for the recognised list in writing before you choose a venue.

Three patterns are common: Web of Science SCIE only, which excludes ESCI; any Scopus-indexed journal, which opens most of this page; and a CORE-ranked conference list used alongside journals, which is specific to computing. A paper that satisfies one can count for nothing under another.

What does Scopus discontinued mean for a paper already published?

Your article stays in the database and remains a Scopus-indexed paper if it was published while the journal was still covered. Coverage ends from the discontinuation date forward, so later articles in that journal are not indexed.

Keep a dated screenshot of the journal's Scopus record from around the time you submitted. It settles the question quickly if an assessment committee later finds the title on a discontinued list.

Which machine learning journals give a first decision in under two months?

No index publishes verified turnaround times, so any list ranking journals by speed is guessing, and we will not print one. The reliable method is to open three recent articles in a candidate journal and read the submitted, accepted and published dates printed on the first page.

Structurally, the faster routes in this field are the open access titles that publish continuously, such as the MDPI, Frontiers and Elsevier open journals above, and the community venues with rolling review such as Transactions on Machine Learning Research. The senior SCIE titles and the subscription journals are consistently the slowest.

Why has my machine learning paper been with the editor for six months?

Almost always because reviewers declined and the editor is still looking. The reviewer pool in this field is stretched thin by conference deadlines, and around those deadlines invitations go unanswered for weeks at a time.

Send one polite status enquiry at three months and another at six. If the answer is vague twice, withdraw formally in writing before submitting elsewhere, and never submit to a second journal while the first still has the paper.

Which machine learning journals charge no article processing fee and are still indexed?

As on October 2026 the no-fee indexed options here are the Journal of Artificial Intelligence Research, Big Data Mining and Analytics, Transactions on Machine Learning Research, the International Journal of Interactive Multimedia and Artificial Intelligence, and the Advances in Distributed Computing and Artificial Intelligence Journal. Three of those carry a Web of Science listing as well as a Scopus quartile.

Beyond those, every subscription title on this page also charges no APC, which makes machine learning one of the cheapest fields to publish in if you can wait.

How do I check whether a machine learning journal is still indexed today?

Take the ISSN from a published article and search it on the live Scopus source list and the Web of Science Master Journal List, then check the discontinued and on-hold files as well as the active ones. Do it on the day you submit, because a list from last quarter is not evidence about today.

Search by ISSN rather than by title. This field has more near-identical journal names than any other in this series, and a wrong match is the first step towards paying a cloned site.

What editors screen before a reviewer sees your paper

The editorial office checks whether the English is clear enough to review, whether the structure matches the journal format, and what the similarity report says. In machine learning a fourth check has become routine: whether a preprint exists, and whether the submission declares it. Posting to a preprint server is allowed by almost every journal here, but failing to declare it looks like concealment. Run your own overlap check first with the Plagiarism & AI Similarity Checker, and settle language and formatting through Manuscript Editing rather than after a rejection.

Where Research Ramp comes in

We match a finished abstract against live indexing, tell you plainly when a design will not survive the first screen, and prepare the submission so the editorial office has no reason to bounce it on formatting. Subject-matched PhD editors, milestone payments, and the author stays the author throughout. Where a paper needs a domain co-author to make the application credible, that is Collaboration Matching, and past work is on our Case Studies page.

Questions we are asked about index tiers

Send us the abstract, we will send back the shortlist

Send us your abstract and we will return a shortlist that separates SCIE, SSCI, AHCI, Scopus and EI options, with the access model marked against each. Tap the WhatsApp icon, or scan the WeChat code. To hand over the whole route from venue choice to acceptance, that is the Full Publication Package.

WhatsApp WeChat

×

Connect with Research Ramp on WeChat

Scan the QR code to add us on WeChat.

Contact page

发送摘要，即可获得按检索类型分层的期刊清单。

Disclaimer. Every table on this page is correct as on October 2026 and no later. Article charges shown in US dollars are approximate conversions at October 2026 rates, and journals bill in their own currency. Indexing, quartile, access model and charges change through the year, and journals can be added to or removed from Scopus and Web of Science at any time. This page is general information only. Research Ramp Private Limited accepts no responsibility or liability for any decision made on the basis of it, including the choice of journal, any charge paid, or the outcome of any submission. Every detail must be independently verified by the author with the journal's own website and the live Scopus source list and Web of Science Master Journal List before submitting. If you would like an informal second check, send us the title and we will look at it for you.

本页所有列表数据截至 2026 年 10 月，仅供一般参考。Research Ramp 不对任何依据本页作出的选刊、付费或投稿结果承担责任。投稿前请务必通过期刊官网及 Scopus、Web of Science 实时数据自行核实收录情况与费用。